I maintain that in the future, any person wishing to learn any skill (not just coding!) will need to willingly eschew the use of AI when learning until they have "built the muscles". The literature is clear that repeated, hands-on practice is really the only way to build skills.
I suspect the progression will be "No AI until intuition (whatever that is for that skill)" -> "Gradual use of AI to understand where it falls short" -> "AI native expert".
How to actually implement this at scale is still TBD ;-) Ironically, AI will be invaluable for this e.g. as a hyper-personalized tutor but it will also present an irresistible temptation to offload the hands-on practice. We already have studies indicating the former is helpful but the latter stifles mastery. At this point I can only see self-discipline as a mechanism to willingly avoid AI.
Unfortunately, our testing-oriented education system only serves to incentivize over-reliance on AI (Goodhart's Law etc.) None of our current institutions and processes are suited for what is already happening and will only accelerate from here on. Things will need to change radically.
For this reason, I once predicted apprenticeships will be a thing again, and already there are signs with Microsoft's preceptorship proposal: https://dl.acm.org/doi/10.1145/3779312
This is highly encouraging because a tech giant is not only acknowledging the problem, but proposing a solution. Not a complete solution by far but at least a start.
Junior devs have always been useless. You used to give them tasks that take them a week or two even though a senior engineer could do it in a couple hours, not because you wanted them to contribute, but because you wanted them to learn to contribute.
The same ethos makes sense with AI, it's just that every company is trying to avoid paying that training tax. Why turn a junior into a senior yourself if you can get the competition to pay for it instead.
It's interesting that the same dynamic is playing out on a much larger scale with children. A child is far more helpless than a junior engineer - at least a junior engineer can feed themselves, wipe their own butt, avoid destroying the room, and generally keep themselves alive. Everybody wants to offload the cost of raising children to parents, because the economic benefits aren't realized for 25+ years yet the costs are very substantial (frequently, at least one parent's full-time attention, costing them an income). Prospective parents are saying "fuck that shit" and simply choosing not to have children.
The long-term effects are going to be much like the effect of the software industry turning away from juniors: total collapse. When you have no workforce, you'll do no work - hell, there is just...nothing, nonexistence, no consumers either. But the fertility bust operates on a longer timescale (I think the software industry will start feeling the dearth of juniors in ~5 years, the economy as a whole won't feel the dearth of children for ~5), and it's far more fundamental. Rather than one industry disappearing, all industries will disappear, likely refactored into something that looks far different.
It also reminds me of those ecological predator/prey/locust models that I studied in calculus class, where population dynamics for many species have a tendency to overshoot the carrying capacity of the environment. Each individual in the population makes their own reproductive & survival decisions, but the sum total of them leads to population collapse and a near total extinction, followed by recovery once the survivors find resources abundant again.
Also sort of. Some form of social organization seems to be necessary for humans to function. But humans are also pretty good (well - relatively speaking, it usually seems to require a war or revolution) at changing that form of social organization as technology, population, and environmental conditions change.
I think this is a very likely outcome. We aren't going to get continued population growth next generation; a significant number of the people needed for it will never be born. This is going to have ripple effects across wide swaths of political and economic organization. But you'll have pockets of population that basically barricade themselves off from the wider economic world and insulate themselves from its collapse, and then the people within them, along with whatever form of social organization they happen to adopt.
> The same ethos makes sense with AI, it's just that every company is trying to avoid paying that training tax.
Last time when a junior dev was added to my team I had a similar thought. But then talking with management I was informed that things went beyond just training.
The company had a social responsibility pledge and understanding with the local educational institutions. They had to pledge to be part of the internship and hiring activities every year. The company could not chose to be fair weather friends and try to recruit people only when they saw fit.
The other aspect was cost. A team made of only senior engineers was costly.
The last aspect was leveling up. Unless the company has lots of levels the team might end up lots of engineers at the same level. And with the inverted funnel nature of promotions it meant some engineers might end up waiting years for the promotion.
So, it was better to have teams with some junior, intermediate and experienced engineers. That way costs and promotion flows were controlled.
Now with AI the impact might go beyond junior devs. I see even the intermediate devs being impacted. It is more likely that companies think they can replace say 1 junior + 1 intermediate with 1 junior dev with AI. Or something along those lines.
This kind of flippant approach is equally valid as "just use ai", "let other companies train the juniors", and "don't give promotions just hire new juniors + ai". All of these have obvious problems from their overly myopic viewpoint.
How is that flippant? I went from an “architect” over the entire cloud developmrnt strategy at a 60 person startup making $160K in 2020 to working at AWS ProServe making $220K (cash + signing bonus + RSUs) as an L5 (mid level - no longer there). Do you think I cared about my title?
I said just the opposite - hire fewer seniors + AI and don’t hird juniors.
I’m now a staff level employee again at a 3rd party consulting company working with small and medium businesses (my preference) after taking a detour in BigTech from architectural roles at smaller companies. My quarterly strategic goals were a combination of what the CxO’s/directors defined and what I defined was in the best interest of the company. None of those goals had anything to do with larger societal issues.
I’ve had some who are useful almost out of school. The amount of tickets is always growing and have someone pick up those “when things calm down I swear I’ll address this” tickets is always helpful. If they can’t get anything done by themselves in the codebase then it gets much harder. I do also think that some people have completely forgotten all the context they didn’t have when they started off xx years ago so mentoring is not always very good
What’s the importance of then learning to contribute if they will probably jump ship anyway when they get good enough? Your HR department is not going to give them a market rate raise to keep them - see salary compression and inversion. A junior developer just isn’t worth the investment.
I have never once told my manager “it would be really nice to have a few junior developers. It would really help us get this project done on time”. They do “negative work”.
Yes not having juniors become seniors is an industry problem. But my goal is to reach my company’s quarterly and anual goals - not what’s going to happen 10 years from now.
> I have never once told my manager “it would be really nice to have a few junior developers. It would really help us get this project done on time”. They do “negative work”.
I have. A good junior can do in a week what a senior with domain knowledge can do in a half day, with only an hour of mentoring along the way. This isn’t a great exchange rate per dollar (juniors are cheaper than seniors, but not that much cheaper) — but seniors with domain knowledge are a finite resource, you can’t get more of them for love or money, while juniors are fresh-minted every semester. The cheapest way to shipping may not go through juniors, but the fastest way usually does; and that’s completely ignoring the HUGE side benefit of building seniors “the hard way,” which is still easier than hiring.
And as a senior+ with domain knowledge, with AI I can do the work of two juniors without the communication overhead + do all of the project management, dealing with stakeholders, etc.
But you don’t build seniors, you build capable mid level ticket takers who jump for more money at the first opportunity.
How is that true? As an architect when I was working at product companies, the director/CxO wasn’t going to call the junior developer to the carpet for not getting a task done or even the mid level ticket taker. Hell they don’t even know what the individual tasks are. I’m going to be the one ultimately responsible either way for project success.
Even when I was working at AWS as a mid level/L5 in the Professional Services division (lower title/lot more money), I was the one who was responsible for my “workstream” on larger projects. I couldn’t say “that’s not my fault. Blame the new L4 junior consultant who just got out of the internal boot camp for new grads”.
Now that I have moved back up to a more senior position in consulting, if a project I’m leading goes sideways, I can’t tell the customer that it’s not my fault, it’s the fault of the workstream leads and the workstream leads can’t tell me, it’s the fault of the more junior consultants who work under them. They will never talk to the client. The people under the workstream leads may not even speak English.
And that’s not meant to be an insult. The workstream leads have to he able to speak passable English and they can work with people under then who only speak Spanish.
Its not about treating them well. The environment and culture isnt the problem. Its that tech companies have decided that they dont like giving appropriate raises to current employees. They are somehow fine with paying the same or more money to bring in an equally qualified external hire. But not with retaining people. Idk why, maybe some stupid MBA rule, mayne theres some good reason behind it
The problem is that (hypothetical) you as a line level manager don’t control comp and raises. Even in BigTech your manager doesn’t control your promotion and everyone knows it’s better to “boomerang” because you will get paid less being promoted to an L5 (mid) from an L4 than someone hired as an L5.
> But my goal is to reach my company’s quarterly and anual goals - not what’s going to happen 10 years from now.
Your benefiting from the work of peopke who did worry about what will happen 10, or even 20 30 years down the line. People like you are why the rides gonna stop
It’s telling that OP worked at Amazon because this pretty much sums up the culture there - new grads churning through and burning out in two years while a few seniors stick around and perpetuate the cycle.
I was 46 when I was hired and it was my 8th job out of college. I was far from a naive college grad. I went in with a purpose - keep my head down for 4 years through my initial four year package, put it on my resume, build a network and leave for a smaller consulting shop.
In fact the only reason I got my job there was because a recruiter reached out to me about an SDE job on the retail side. There was no way in hell that I was going to sell my big house in the burbs of Atlanta (at the time), uproot my life after Covid and work for Amazon. I knew what I would be getting into.
She then told me about a “permanently remote”,”field by design” role at AWS ProServe. I was like sure why not?
>What’s the importance of then learning to contribute if they will probably jump ship anyway when they get good enough? Your HR department is not going to give them a market rate raise to keep them - see salary compression and inversion.
Obviously that hasn't historically been true, else there wouldn't be any senior developers as companies would have wised up to that and nobody would hire them as juniors.
- Not everybody is a job hopper (even in Silicon Valley one sees that most junior FAANG devs stick around for a good while).
- The HR department is absolutely going to give junior developers that pass the cut after a year or so a market rate raise.
- In limited hiring periods, they'd be grateful to have the chance to stick around, while in bullish "boom" periods companies can afford to spend to keep people, expand and give them bigger roles, and so on. It's in the in-between that it becomes more problematic, but now we're in a "limited hiring" era.
>Yes not having juniors become seniors is an industry problem. But my goal is to reach my company’s quarterly and anual goals - not what’s going to happen 10 years from now.
That's how companies fail.
It's also not a good strategy at the personal level. If you command more devs, you get more leverage.
This is not true - the average tenure for a developer across the industry has been 3 years for well over a decade.
> The HR department is absolutely going to give junior developers that pass the cut after a year or so a market rate raise.
This is also not true from small companies to FAANG - see “salary compression and inversion”
> That's how companies fail.
The company failing in the long term is really not any current employees main concern unless you are a founder if the average tenure is 3-5 years. Even the stock market doesn’t care about the long term viability of a company.
BigTech for instance can afford dead weight. Amazon has an internship program and for those who come back or through their non traditional programs for their internal consulting division (AWS Professional Services) they have a 3 (6?) month training program.
In ProServe at least (former employee) even for their l5/L6 employees, they have the 3 month training program - “AWSome Builder” where you simulate a customer project and have to pass.
After leaving AWS and being hired as a staff consultant by a third party company, they put me on a plane two weeks in to meet with a customer. They don’t even hire less than senior+ people in the US.
If the median tenure is 3 years and the software business still is very profitable then people must be net useful within that 3 years. A lot of people also just don’t want to job hop much and honestly the interview culture keeps me from hopping more. I do still fall in the 3 years per hop but I’ve always had a good reason to- ie. layoffs, company going in a worse position than when I started, shit management at various levels, forever compounding responsibility…
>This is not true - the average tenure for a developer across the industry has been 3 years for well over a decade.
That counts temps, people who weren't a good fit and were let go early after hiring, mass layoffs, and mixes mixes startups and FAANG and consulting churn, none of which is the typical corporate IT worker scenario, and all of which bring the average down (but are not "hopping").
Corporate IT, government IT, smaller SMEs, and stable SaaS, have higher averages.
It doesn’t make sense to hire juniors at all other than as a service to society. I haven’t hired a junior in 4 years. The one I hired 4 years ago was because not only did he do reasonably well on the interview but he literally begged me because he trained himself to do it while painting houses so I saw a lot of passion in him.
So exactly what is the point for a profit seeking company to train someone except thsr you expect them to bring you more business value than they cost during their tenure?
This is the difference between being an engineer and being a clock puncher. You don't care about the business, you don't care about the product, you don't care about society as a whole. So long as you get your paycheck and your annual pay bump, fuck absolutely everyone and everything else, right?
Don't worry, just leave all your problems for someone else to fix. I'm sure that won't have any lasting consequences at all.
I work for one reason and one reason alone - to trade 40 hours of labor for money to support my addiction to food and shelter. The company is not going to give me money for “caring about society”. They are going to give me money to meet my quarterly goals to help them meet their profit goals for the company and in a former life , to make them look good for the public market pre-IPO and at another company for an acquisition.
I give a company 40 hours a week and all of my 30 years of industry experience and they give me money (and in a former life RSUs)
Sweet summer child. I was once opinionated and driven as you are now. I remember when I got out of college, I also thought like that of the mediocre clock punchers.
Now at my 45 years, I couldn't care less for whatever grand objective the current company I work for has. I exchange my knowledge and time for hard cash, and let the owners , ceo and whatnot run with their grandiose vision.
Funny enough, the more I got into this mindset, I slept better, made more money, and got more autonomy.
Once directors and CxOs know that you are completely aligned with the business goals and ignore everything else that “doesn’t make the beer taste better”, they trust your judgement and basically leave you alone.
More money for me until I retire - well actually I’m past the point where I chase more money - and then after that - it’s not my monkey and not my circus.
Again HN bubble thinking. Most developers in the US are working at banks, airlines, insurance companies, etc in second tier cities - in the “enterprise” and are not making “$400K”. Most developers will never in their career see more than $175K inflation adjusted and I really haven’t seen comp on the top end increase in nominal terms in a decade [1] for enterprise devs.
That leads to my second point, in second tier cities, you see comp go from around $80K —> $115K -> $150K —> $175K, junior -> mid (pull well defined tickets off a board) -> Senior (leads larger initiatives) -> Senior+.
For instance look at what Delta airlines pays based on Atlanta.
Agreed. We are still in a capital crunch so overhiring is out of fashion. People don’t remember the early 90s or the dot.bust when the same things were said.
Kraft 1977 Programmers and Managers talked about this if I recall. Still the best alternate take on our industry I have ever read.
Yep. This is why many companies have a terminal level with “up or out “ rules. Before that level you are not fully independent and require too much supervision. No one wants a Jr engineer with 10 years of experience.
I see a lot of Sr engineers get very frustrated by how much time they have to spend helping Jr engineers. But, that’s the job, or at least a big part of it.
I burnt out helping a junior on my team for the past few months. It was just terribly obvious she was feeding my responses directly into a chatbot to fix instead of actually understanding the issue. I can’t really even blame her, there isn’t much incentive to actually learn
I've been in situations like that. For me, it's like interviewing, I just keep backing off, lowering the bar, making it easier and easier until they can get it, then start going back up again. I pretty quickly get a confident read on where they are.
If at that point it's clear (to me) the situation is not salvageable, it's a management issue, I've done my job.
Strongly disagree with this. Bad junior devs might be useless, but I’ve seen good ones absolutely tear through features. Junior devs fresh out of school typically have tons of energy, haven’t been burned out, and are serious about wanting to get work done.
And it’s not a waste of their time to have to give detailed requirements and troubleshooting steps to a junior developer, constantly being interrupted, and then having to check their work thoroughly?
If you have to be that detailed anyway - you might as well use AI.
So exactly how am I going to convince my management to open a req for a junior developer who is not going to help us meet our quarterly goals and take time away from the other senior developers that will either have to work longer hours or do less work?
I’m not going to work as a charity and neither are any of my coworkers. We are all here to exchange labor for money.
We as a collective need to convince our management of this, but that needs to start with people getting their heads out of their asses and working together instead of this mercenary attitude you have
I don’t have to do anything except keep my head down, do my job and enjoy my well earned autonomy. I’m definitely not going to try to convince my skip, skip, skip manager to change their hiring policies. It’s not like my line level manager has any power over anything
Even when I was at a startup before 2020 and I did have the ear of the CTO and the founders I knew my ultimate mission was to do what was needed to get acquired and before that I knew exactly what my mission was when k was hired to lead the tech initiatives as we were acquiring companies “find efficiencies” and go public.
Or do you think I could have convince anyone of anything as an L5 at AWS in the middle between architect at a startup and my current company?
I’ve gotten plenty of use out of junior devs. The critical bit is what makes anyone a useful worker. I’ve found anyone that’s both dedicated and meticulous is worth the investment.
Sure there’s a wide range of skills and you can’t just hand any task to anyone and expect it to work out but some fresh collage graduates are more capable than the average person with 5 years of professional experience. At the other end you need to focus on whatever they actually are capable of doing. 40+ hours a week can slowly expand even an extremely narrow skillet as long as they’re a hard worker.
You don’t want 3 Jr devs at the same time because of diminishing returns. Most projects have grunt work where attention to detail is important but experience doesn’t really help much. AI can quickly come up with alt text for images, but ensuring it’s actually useful for someone using a screen reader is a different story.
1 new Jr every 2 years works quite well for a team of 7+ developers.
And that junior dev will probably leave in 2 years because HR won’t allow you to give them a raise to match their market comp.
But with a team of 7, I can’t believe that giving them all a Claude Code subscription wouldn’t be much cheaper and much more productive than hiring a junior. A junior though with AI is more dangerous than a junior without.
Working on problems that are a poor fit for a mid level developer with AI assistance of which there are still plenty.
I find AI tools make everyone better at exactly the same kind of problems, which means a larger percentage of what’s left over is more cheaply done by Jr’s working without AI assistance.
A mid level developer is just a ticket taker. That’s their job. No matter what their title, if they are just pulling well defined tickets off the board, that’s what they are.
How exactly do some people leverage AI to rearrange the furniture in the conference room for an office party?
Physical labor may be a tiny fraction of what a team of developers do, but I’ve seen what amounts to several thousand dollars spent on that kind of silly task because teams leverage the tools they have.
You just replaced one task that an AI is useless for with a different task that an AI is useless for.
The point obviously still stands, and no I am not suggesting using Jr devs for physical labor alone is worth adding them to the team. Rather that “Work” includes a very wide variety of tasks that need to get done.
> I find AI tools make everyone better at exactly the same kind of problems
this part I find fascinating. most places I worked (30 years in, 10 as a contractor so quite a lot) the distinction between Jr, Mid and Sr falls exactly into the kind of works that they do. I, as a Senior, often work on hard shit, Jrs are not (yet) entrusted with those problems and work on entirely different set of problems. I cannot compute how AI makes everyone better at exactly the same kind of problems, problems I am solving today few people on my team are working on and then same goes down the "pyramid"
> Junior devs have always been useless. You used to give them tasks [...] not because you wanted them to contribute, but because you wanted them to learn to contribute.
Junior devs are by your own explanation not useless. They are the most important human investment in your project.
I mean, if we’re doing this, let’s be honest and go as far as mid-level engineers whose work needs constant correction, as well as the many, many senior engineers out there who are senior only because they lucked out in getting the title during the artificial dev scarcity of the ZIRP eras.
I think the idea of 'junior' needs to be refined a bit. By the time I got my first job I've been coding for years, and have built rather substantial things.
In fact, in terms of pure coding ability, I was probably past the initial, fast part of my growth.
As should have others, which the university education system should have made sure.
The fact that some people come out of 4+ years of software engineering education utterly clueless means that they somehow managed to dodge having to build anything, I think means that they will never get good at any point in time, as they either were very talented at dodging having to build things, and I don't think that talent is going to abandon them, or they couldn't really grasp the basics in an environment designed for just that.
With that said, I think you can see for most juniors, what you can expect out of them in terms of pure coding ability - sure a lot of them have room to grow, but I've met so many great people who were very young, yet were useful from day one.
In fact, if you have the willingness to grind away at some problem, that puts you ahead of a significant amount of the pack. I have had the misfortune of working with people who lacked any demonstrable skill, and had coping strategies for having to deal with any sort of hardship. Getting useful work out of them was a challenge in of itself.
These people managed to get the years in to be considered senior, and are probably dispensing their wisdom 'mentoring' juniors somewhere else, and are no longer expected to actually contribute to meaningful issues.
I'm not sure if enthusiasts are the exception rather than the norm? I've noticed in the last few years, a lot of junior engineers do not have much active coding experience outside of their university education, they aren't the traditional "obsessed with computers and programming as kids".
There has been a much higher demand for software developers over the past 10-15 years than there are people who are obsessed with computers and programming
If you look at the general topic shift on HN over the years it's obvious most people are getting into tech because they want power and money, not for love of tech
> If you look at the general topic shift on HN over the years it's obvious most people are getting into tech because they want power and money, not for love of tech
I think that is a bit presumptuous, even if there is a hint of truth there. Just because a kid didn't grow up around computers and programming shouldn't exclude them from the tech path or mark them as imposters/pretenders.
> Junior devs have always been useless. You used to give them tasks that take them a week or two even though a senior engineer could do it in a couple hours
We havent dont it and I never seen something like that.
> > For someone like you, who likely has years of experience without LLMs, your brain totally understands good code/bad code, good architecture, and just general intuition around code and systems. LLMs must be an absolute gamechanger. But for someone like me who is starting out in this field, how am I supposed to build the years of experience and intuition that comes from manually writing code and building systems when companies are expecting AI to be used from here on out?
Answer: juniors need to work with seniors, and the seniors need to teach the juniors, and the juniors need to learn to use LLMs to learn, not just to do the work.
Next year I'm going to experience my first junior, which is crazy considering I have 3 YOE. Nonetheless I'm very curious what kind of graduates come out of the current education
It's a self inflicted wound. Companies do not reward loyalty. They do not give out raises congruent with what you can find if you leave. Business-types unirionically think seasonal layoffs is a "good thing." Self hemorrhaging your institutional knowledge is insanity
1) People hearing "an LLM is as smart as a junior" and actually opting for the LLM subscription price instead of hiring a junior
2) The gap between senior and junior in terms of performance has become larger, since the senior devs had their hands get dirty for years typing stuff out manually AND also tackling challenges.
This generation of junior-mid developers will have a significant portion of the "typing stuff" chopped off, and we're still pretending that this will end up being fine.
I think your second point is interesting, and it has actually already happened a couple of times.
It used to be a lot easier to find devs that knew assembly and could navigate call stacks through memory by hand because a lot of folks had to learn that to get their job done. Now higher level languages have mostly eliminated that level of operation.
The same applies to infosec roles. It is 10x harder for junior infosec folks than 20 years ago because there are a bunch of skills you need in infosec that today's mainline dev experience doesn't need, but were more common a while ago.
Case in point, I remember working with a partner company's junior engineer on some integration. They needed some hard-coded constant changed and time was of the essence. I told them to change a couple bytes in the elf binary directly. They looked at me like I was a wizard. I thought it was a fairly pedestrian skill having grown up reversing computer game save files.
The challenge is to get cost sensitive businesses to support this. Juniors are a cost and when trained move on, thats the fundamental problem. Retention only works with smart companues, for most other companies its a revolving door.
On the plus side, as a dev with 30+ years of experience, I am commanding a very good contract salary these days. Revolving door companies stuck in process hell and product rot, and cannot deliver new value, so they’re scrambling to find experienced devs that cost a premium. My salary today makes up for peanuts at the start of my career.
The real question will be; Do we need to pay the juniors to write code to become seniors?
If coding is an art then all the juniors will end up in the same places as other struggling artists and only the breakout artists will land paying coding gigs.
I'm sitting here on a weekend coding a passion project for no pay so I have to wonder.
> Seniors come from juniors. If you want seniors, you must let the juniors write the code.
Companies know this as well, but this is a prisoner dilemma type situation for them. A company can skip out on juniors, and instead offer to pay seniors a bit better to poach them from other companies, saving money. If everyone starts doing this, everyone obviously loses - there just won't be enough new seniors to satisfy demand. Avoiding this requires that most companies play by the rules so to say, not something that's easily achieved.
And the higher the cost of training juniors relative to their economic output, the greater the incentive to break the rules becomes.
One alternative might just be more strict non-competes and the like, to make it harder for employees to switch companies in the first place. But this is legally challenging and obviously not a great thing for employees in general.
It’s already getting harder to find juniors willing to write the code and harder to discern whether someone is as willing as they say. And I feel like asking junior to make this decision and just have self control is a tricky double edged sword. Even if I want them to (and I do!) the competitive and ambitious juniors I suspect will still lean into AI code gen heavily as it makes them look better and seem more productive. Seniors probably need to do more than let them write the code, we probably need to figure out ways to encourage, require, or even enforce it at some level, if we want it to happen.
I agree with the sentiments here. But, I’m less hopeful about the presented solutions.
I think my argument against humans still needing to know how to manage complexity, is that the models will become increasingly able to manage that complexity themselves.
The only thing that backs up that argument is the rate of progress the models have made in the last 3 years (ChatGPT turned 3 just 3 months ago)
I think software people as a whole need to see that the capabilities won’t stop here, they’re going to keep growing. If you can describe it, an LLM will eventually be able to do it.
disagree because when the "super fast" new CPUs of 20 years ago became common, it was easy to write code that executed slower than previous code, due to language constructs and wasteful work patterns. Therefore, I predict that LLM code can explode in complexity (14KLOC for a binary file parser with some features) but that compute will bog down and effort to understand will explode.. that is, in extreme cases.
Ok but even pre ai I felt like each years interns wanted to take as many shortcuts as possible and not learn.
I think the allure of high TC (150k base or more for entry level) led to many non engineer brained people to enter tech.
Many people can do rote memorization, it’s even ingrained heavily in some cultures iykyk. However they can’t come up with much original or out of the box thinking.
The silver line is not to avoid AI but to use it wisely, and this is coming from somebody who used to hate AI.
I am not a Python developer or developer for the matter but Perplexity AI did help me to understand the bsic of Python for API requests and get projects delivery with 94% code coverage and vulnerability free.
AI also reduced the time spent with Ansible playbook generation, but I do know Ansible, I do know Linux, homelab is my hobby so I am not just doing copy and paste. I review whatever it generates and correct it when required.
In the companies I have worked and work, I see developers themselves confessing "I used AI, it works but idk how"!
AI itself does not make you useless, only and only if you used it as a smart search engine.
If you are doing copy/paste, you are going to get so screwed professionally speaking.
Folks are no longer learning and what they are doing, AI can do on its own.
My nightmare scenario (which might start to materilize) is that our last years in the industry will be becoming prompt monkies / agent "managers" working on codebases we barely understand in such velocity there's no way we can gain real understanding. Whenever something breaks (and it will , a lot) A.I will fix it - or so we'll hope. And the sad thing is - this might work; you'll get more stuff done with fewer people. Sure, we didn't sign up for this, it's not a fun job what I've described, but why should management care? They have their own problems and A.I is threatening their jobs as well.
At work we build enterprise software with stuff like Kotlin+Spring + multiple NextJS apps + Microservices + Rust CAD engine.
I haven’t have written code aside from tweaking stuff here and there in probably 3 or 4 months. Before that I wrote code by hand every day for many years.
I’ve found a lot of fun parts of my new workflow that I enjoy. I still miss being fully immersed in a problem deep in the files… and sometimes it feels like homework reading so many implementation summaries from Claude because the feature spans 4 repos and is too much code to read. But I do love shaping the code into different solutions exploring in a way that is unique to ai native workflows. And I love building agent skills and frameworks with/around them and expanding it out to more aspects of the company or life — there’s deep work to be had that still feels like hacking in the trenches. I get a lot of the same satisfaction in different ways, and there’s a lot of exciting novelty to explore that was previously out of reach due to time and energy constraints.
Also I don’t like our backend stack and I hate React / NextJS to the degree of derangement syndrome — I am so happy that I don’t have to write it and I can just focus on UX, making customers happy / lives easier / shaping the software into better and better versions of itself at such a faster pace.
People who learned good software engineering intimately before the inflection point are extremely lucky right now. Existential dread and the stages of grief have been a part of the journey for me too sadly, but there’s a lot to celebrate and explore with the right attitude.
Similarly, I started using Claude to add some features to an native app on iOS and Android in early January. That was so successful (tooks days instead of weeks) I started applying it to client work and basically haven't written any substantial code since. A big change from around 40 years of writing code pretty much every day and I'm enjoying the increased velocity from not having to web search for API and CSS syntax details.
My son, working in another dev company, reports the same - he basically hasn't written much actual code for about three months. It's a massive change.
I feel the same way, I have many years of experience, and I have gone from writing everything by hand to using claude code all the time (my latest company is very pro doing everything with AI).
Since I have been a software architect for the past 7-8 years it feels in some ways that that experience makes using claude code a lot more productive than for my non-architect colleagues, as I am able to steer it much more effectively whether directly in sessions or via custom skills / mcp.
The big issues right now for me are hiring and manager expectations, I changed positions last fall due to mass layoffs and it took me 3 months to find one: having leetcode interviews in the current climate seems completely useless, even more than it was in the past, and system design interviews are so formulaic it also feels like a crapshoot. Plus every job getting hundreds of AI generated applications makes actually being considered in the first place quite difficult.
Manager expectations are also ridiculously inflated nowadays, it seems most action items that come are claude written with fantastical random statistics (if you add caching you can make your backend 98.3% faster!), and it takes so much time to fight this and unrealistic team velocity expectations.
Interesting times, I do feel lucky I have had a long career, but I very much fear the ladder being pulled up even more than it has been when outsourcing because widespread. I know everybody says "things always change, new opportunities will open up to compensate for the ones that are being lost" but this time it does feel different, and not in a good way.
Things are changing so fast and so chaotically with this technology. I'm also writing everything now using Claude code, and I've been thinking a lot about what this means for my work moving forward. One thing I've noticed, is that I will just keep hammering and hammering on my work until I force myself to quit. Even on the weekend I feel the pull to go work on it. I'm just less sort of mentally exhausted by work, I suppose, but I don't think that's particularly healthy if it leads to me working way more than I should. On one hand, I think that's a reflection of how powerful and exciting this technology is, but on the other hand, I think that it triggers some different kind of reward function in my mind that I'm not used to.
In any case, I think if one wants to continue to have a career in this industry for years to come, it's basically table stakes to become fluent in using these tools.
It already happened. The old timers correctly observe that modern applications are bloated and inefficient because of all the heavyweight frameworks, excessive abstraction layers, and "left-pad culture" where external dependencies are pulled in to do the most trivial things but that these things enabled less capable developers to effectively build software to fulfill industry demand. LLM-only coders are just the next step in the devolution.
> My nightmare scenario (which might start to materilize) is that our last years in the industry will be becoming prompt monkies / agent "managers" working on codebases we barely understand in such velocity there's no way we can gain real understanding.
It will always be preferable to work on an understandable codebase, because that maximizes the AI's affordances too. And then the AI can explain things to you. A skilled human will always have a lot of solid knowledge relating to their hyper-specific niche that isn't part of your average general purpose AI, so humans will obviously have a key role to play still.
I'm already seeing this in the company I recently joined: 80-90% of code is generated/prompted. Big PRs, very little review or oversight. Absolutely nobody considering long-term architecture (and IMO nobody capable of such). In general, there's very little critical thinking involved at any stage, just throw error messages back into the LLM, rinse and repeat.
I'm hoping there's a world where people with skills are useful in getting these projects back on track, but perhaps as a society we're learning to accept this reduction in quality.
And how do u sum up the tradeoffs so far, or is it too early to tell ? Do u see lots of unacceptable shit making into production that wouldn't have before A.I for example ?
This is 100% an issue on the side of the senior developers. Imagine saying "these juniors are useless" because you are making them work in assembly, but C has just been released. You are giving them menial work that is no longer required to do by humans. Instead of giving them the task "update these email templates", the norm should be: "create this new service that automates an internal process". They will make mistakes and they will learn - but what they will be doing is going to be very useful and also give them chance to grow the necessary skills for this new era, with the supervision of a senior.
By making them walk into the landmines and forcing them to fix it - that is how everyone became a good programmer. It's just the scope that has changed.
Actually the truth is that a lot of senior devs are not very good either, and have negative value. But they have an inflated value of themselves that does not reflect reality.
Pretty much all software projects seem to peak, and then decline in quality. There are only a handful of senior devs in the world who are actually good programmers.
I agree. But it’s not about that they have inflated value but it comes down to how “modern” software producing organizations work. Product managers and C-level people do not know what they are doing either. Most people are part of the “software engineering” theater, recruiters are recruiting, manager are managing, software developers are developing software, all of them just to get paid or gain status in the org. Most of the values come from that handful people who can really deliver.
> The mechanistic myth is the belief that
everything can be described as a neat hierarchical
structure of things within things. And few of us
realize that our entire culture is based on this
fallacy. While the world consists of complex,
interacting structures, we prefer to treat every
phenomenon as a simple, isolated structure.
> Through our software pursuits, the mechanistic
myth has spread beyond its academic origins, and
is now affecting every aspect of human existence.
In just one generation, it has expanded from
worthless theories of mind and society (behaviour-
ism, structuralism, universal grammar, etc.) to
worthless concepts in the field of programming
(structured programming, object-oriented pro-
gramming, the relational database model, etc.)
to worthless software-related activities that we
all have to perform.
> What is worse, our mechanistic beliefs have
permitted powerful software elites to arise.
While appearing to help us enjoy the benefits of
software, the elites are in fact preventing us
from creating and using software effectively. By
invoking mechanistic software principles, they are
fostering ignorance in software-related matters
and inducing dependence on their systems.
> Increasingly, in one occupation after another,
all we need to know is how to operate some
software systems that are based on mechanistic
principles. But our minds are capable of non-
mechanistic knowledge. So, when the elites force
us to depend on their software, they exploit us in
two ways: by preventing us from creating better,
non-mechanistic software; and by preventing
us from using the superior, non-mechanistic
capabilities of our minds.
> The ultimate consequence of our mechanistic
culture, then, is the degradation of minds. If we
restrict ourselves to mechanistic performance,
our non-mechanistic capabilities remain un-
developed. The world is becoming more and more
complex, yet we see only its simple, mechanistic
aspects. So we cope perhaps with the mechanistic
problems, but the complex, non-mechanistic ones
remain unsolved, and may eventually destroy us.
Wow, I'm immediately hooked. The book is free online as well to download.
It feels like to me that junior devs don’t understand what they even need to learn. They just use agentic coding to get things done, without any deeper knowledge.
The worst is, they think they know exactly what they need to learn, and also think they can make good decisions.
I can't seem to get the article to load, but I think I get the gist from the title.
I hired a junior "dev" who literally hadn't even completed an HTML course. Before AI I could not have hired them because they literally did not know how to dev. After AI, anyone with a little grit can push themselves into the field pretty easily.
As with everything in life: you can choose to hard route or you can choose the easy route and your results will follow accordingly.
Adam, can you please share, how in the world, this junior dev got hired with you?
I'm self-taught dev with multiple years of experience. I choose the hard route, even after AI. For me, programming is theory building, so I always choose understanding above all else.
Rock solid understanding of TypeScript, frontend and backed.
I have sent 100s of CVs. For Juniors, Mids and Seniors. Not even a single interview.
From a professional standpoint, I’ve never found “enjoyment” from coding. I enjoy every part of the process of getting business goals from talking to stakeholders -> completed project with coding being the necessary evil.
Funny enough, I started working in 1996 professionally (and had been a hobbyist for six years before going to college). But it was only between 2012-2016 that I was a ticket taker without working with the end user directly - everything I’ve done has been B2B.
GenAI (and working remotely since 2020) has made me enjoy every part of my job.
> I hired a junior "dev" who literally hadn't even completed an HTML course.
I mean, I'm a fairly senior dev, and have literally never completed, or indeed really heard of, a html course. Is that, eh, part of your average CS degree these days?
> Netscape has stipulated removal of the <marquee> element from the Internet Explorer during an HTML ERB meeting in February 1996, as a condition to removing the <blink> element from the Netscape
It's like nuclear disarmament treaties, but for annoying things.
I can't wait until the AI people realize that without developers' original ideas, AI has nothing new to steal from. We don't create, AI will spit out the same old concepts. What, you're gonna create the next generation of AI by training it on what the very same AI has already produced? C'mon now.
You don't get technical creativity reflexes by using AI. This is technical stagnation in the making. By cannibalizing its own sources, AI is ensuring that future generations are locked-in subscription models to do the most basic technical tasks. This is all obvious, yet we speed up every chance we get.
It might be a mistake to assume tomorrow’s training looks like today’s. Unsupervised learning is a thing and a very hot research topic, precisely because it avoids some of today’s big problems with acquiring the vast amounts of training data necessary.
Unsupervised leanring has been around for years and is already how the current wave of models are trained. It doesn't mean no data, it means no human provided labels of the data. So you still need creative new human ideas to move LLMs forward. LLMs != intelligence.
Exactly this. Even frontier models like Opus 4.6 have absolutely zero understanding of the task at hand. If you give them problem they have not encountered in training data, they will not solve it. You can however, guide them to resolve the problem, but in that case these get reduced to merely an auto complete. Don't get me wrong - models are getting better and hide very well that they don't understand anything, you can almost get fooled now.
Maybe there aren’t that many new/necessary ideas that can be mined from the fundamental building blocks of software development (languages, syntax, runtimes, static analyses, type checking, etc). Maybe people will continue to innovate by instructing models to build novel things out of those building blocks? Perhaps things we would not have thought of building before due to the effort required without LLM assistance.
It's the opposite. The less competent the average developer the more valuable coding LLMs become (as the only way for those bad developers to generate ok code). Eliminate the good developers and even bad coding LLMs become valuable.
I opened LinkedIn today (out of habit) and the first post was someone explaining how much Slack costs and that with AI every company can build their own Slack for $100. So that person decided to build an open source Slack clone using Claude Code. Granted, these were a few sane comments showing good alternatives that have already been built.
But for me it's been a signal that people have no imagination, so they are just burning tokens for no reason.
This is fine. How else do you learn but by taking things apart and rebuilding them? This obsession with productivity is incompatible with onboarding new talent. Having 1000 versions of the same concept is exactly what progress is.
Why would there be a lack of original ideas? People who are born to code so to speak will do it. Information wants to be free as the saying goes. It only takes one time for an innovation for it to be to copied everywhere.
We don’t need the same volume of developers to have the same or faster speed of innovation.
And conversely if there is stagnation there is a capital opportunity to out compete it and so there will be a human desire to do the work.
Tl;Dr. People like doing stuff and achieving. They will continue to do stuff.
ps it’s too much to claim other people don’t experience creative ideas using AI. You don’t really know that’s true. It hasn’t been my experience as I have had the capability and capacity to complete ideas on my back burner for decades and move onto the next thing.
That’s the big scary point at the crux of all of this - you’ve had decades without the tooling to develop instincts. Nobody knows whether it’s possible to develop instincts with the tooling or what those instincts will look like. Creativity takes a degree of skill to execute on and the concern is that we’re potentially graduating people to painting the ceiling of the Sistine chapel before they’ve even learned to sketch.
At minimum, our current generation of leaders will have to get much better at managing resources and building people up. We have to up our games and build environments where the pursuit of deep understanding is permissible. Unfortunately with the current hiring issues, it’s totally understandable that young developers are scared to take time on tickets.
I can't repair my car, which used to be the hallmark of technical masculine skill in the era of Grease the musical, because mechanical maintenance is not primary in my lifetime. Nor do I have any idea how to manage a farm. I think the kids will be fine. On the other hand, I fear I will not survive once my Internet connection goes out.
Innovation is irrelevant to pushing up this quarter's numbers. No one actually values unique and novel ideas. The only thing that matters is shipping something right now that can make an impact on this quarter's numbers.
Who cares if it's derivative slop or a straight up bootleg of something else so long as the number goes up
It isn’t about training anymore. It is about harnesses.
Just look at new math proofs that will come out, as one example. Exploration vs Exploitation is a thing in AI but you seem to think that human creativity can’t be surpassed by harnesses and prompts like “generate 100 types of possible…”
You’re wrong. What you call creativity is often a manual application of a simple self-prompt that people do.
One can have a loop where AI generates new ideas, rejects some and ranks the rest, then prioritizes. Then spawns workloads and sandboxes to try out and test the most highly ranked ideas. Finally it accretes knowledge into a relational database.
Germans also underestimated USA in WW2, saying their soldiers were superior, and USA just had technology — but USA out produced tanks and machinery and won the war through sheer automation, even if its soldiers were just regular joes and not elite troops.
Back then it was mechanized divisions. Now it is mechanized intelligence.
While Stalin said: Quantity has a quality all its own.
I hear this sentiment a lot but it doesn’t ring true for me.
What is an idea really and what’s your definition of new?
If i get a LLM to spit out, I dunno, a deployment system written in haskell that uses bittorrent or something, none of those bits are new, but certainly there will be unique challenges to solve in the code and it’s a new system.
Where is the line for new? Is it in combining old ideas? If not then does any software have “new” ideas? It’s all combinations of processor instructions after all…
While that’s kind of true in some sense, I think there’s an argument to be made for the contrary: that the mechanism for generating new ideas in humans is not quite as special as we would like to think.
In other words, creativity in humans is arguably just as derivative as in machines.
I think this can be falsified by just considering the history of humanity. It wasn't that long ago that human language literally did not even exist. And our collective knowledge wasn't all that much more than 'poke him with the pointy end'. Somehow we went from that to putting a man on the Moon, unlocking the secrets of the atom, and more. And if you consider how awful we are at retaining/sharing information and just general inefficiencies due to the fact that we're humans and not just logical information processing machines, we did all of this in little more than the blink of an eye. This is something that seems to certainly be rather special.
All that humanity has achieved happened due to the simple loop of identifying a desire/need and finding a way to satisfy it. Also known as reinforcement learning. The only thing that really differentiates humans from machines is... history. We've been learning and passing on our knowledge to successive generations over millennia. Nothing really special there; give the machines a few years to learn and see what happens.
I’m not claiming an LLM is structurally or functionally equivalent to a human brain. I just said that what we call “creativity” is in fact a very derivative thing.
What I am excited about is the possibility of LLMs to draw conclusions from the last 150years of scientific papers.
There have been lots of instances of knowledge being rediscovered even when it was previously published but sitting on some shelf forgotten. LLMs ability to digest large volumes of data will I think help with this issue.
We will still need to reproduce and verify conclusions but will be interesting to see what might come from this.
i don't think all sides of this discussion agree on what a "new idea" is. i am a very creative person but i've never had a truly original thought and i don't know how having one would be possible
If AI could innovate it wouldn't be a public product. It would be a cash cow. Why give your customers the ability to come up with new and amazing ideas when you can just keep it for yourself and launch a thousand products? USA is a capitalist society. It doesn't share profitable ideas.
And if AI was really about productivity they'd be talking about doing more faster with the same workforce, not reducing the workforce.
Technically all the problems that almost any given business needs to be solved today has already been solved umpteen times over the years. There are no new problems that can't be solved by porting and/or combining old solutions.
"Everything has already been invented." - Some 19th-century scientist who had no imagination to see the wave of technological innovation that was coming.
That's the literal definition of stagnation. That is not compatible with growth.
Also that's not a new idea, that "everything worth inventing/exploring has already been". It's precisely what AI reinforces, and that goes against human nature (and capitalism) as that statement has historically proved.
I think this concern is overblown. AI is an incredible teaching tool. It's probably better for teaching/explaining than for writing code. This will make the next generation of junior devs far more effective than previous generations. Not because they're skipping the fundamentals...because they have a better grasp of the fundamentals due to back-and-forth with infinitely patient AI teachers.
Unless your company is investing in actually teaching your junior devs, this isn't really all that different than the days when jr devs just copied and pasted something out of stack overflow, or blindly copied entire class files around just to change 1 line in what could otherwise have been a shared method. And if your company is actually investing time and resources into teaching your junior devs, then whether they're copying and pasting from stack overflow, from another file in the project or from AI doesn't really matter.
In my experience it is the very rare junior dev that can learn what's good or bad about a given design on their own. Either they needed to be paired with a sr dev to look at things and explain why they might not want to something a given way, or they needed to wind up having to fix the mess they made when their code breaks something. AI doesn't change that.
I always say "own the output". No need to do it by hand but you better damn well research _why_ the AI chose a solution, and what alternatives there are and why not something else, how it works and so on. Ask the AI, ask a seperate agent/model, Google for it, I don't care, but "I don't know the LLM told me" is not acceptable.
For me, the hardest part of software development was learning incantations. These are not interesting, they're conventions that you have to learn to get stuff to work. AI makes this process easier.
If people use AI to generate code they don't understand, that will bite them. But it's an incredibly tool for explaining code and teaching you boring, rote incantations.
As a junior, my top issue is finding valuable learning material that isn't full of poor or outright wrong information.
In the best and most generous interpretation of your statement, LLM's simply removed my need to search for the information. That doesn't mean it's not of poor quality or outright wrong.
I suspect that the quality is ironically correlated with the expertise of the user (i.e. it is knowledgeable if you are knowledgeable), which puts you in a conundrum (I can report that with a couple decades of experience, LLMs are giving me high quality, correct results, but I can already see that it somehow doesn't work as well for some of my less experienced colleagues. A lot of what I've been doing over the last couple months is trying to find how to make it "just work" for them.).
As a general principle, take advantage of the fact that it can easily generate stuff. If you don't know whether something is true, have it prove it. Make a PoC/test/benchmark to demonstrate what it's saying. Have it pull metrics that you have access to. Add more observability. Create feedback loops (or rather, ask it to create feedback loops). They're very good at reasoning given access to the ground truth, so give them more ability to ground themselves.
They also have fantastic knowledge of public things, but no knowledge of your company, so your instructions should mostly be documentation of what's unique to your company. If it can write an instruction on its own (e.g. how to use git or kubernetes), it is a useless instruction; it already knows that. What it doesn't know is e.g. where your git server is. It also doesn't know what matters to your company: are you a startup trying to find product market fit? Are you an established company that is not allowed to break customer setups? etc. You might even be able to ask it what kinds of questions a senior might ask about how a company/team works when coming into a new job, and then see if you can answer those questions (or find someone who can). In fact, go ask chatgpt:
> What are some questions a senior engineer might ask when coming into a new role to make themselves more effective?
> What are some questions a principle engineer might ask when coming into a new role to make themselves more effective?
> What are some questions an engineering manager might ask when coming into a new role to make themselves more effective?
> What are some questions an engineering director might ask when coming into a new role to make themselves more effective?
This is the dumbest thought that proliferates this website.
Super great that it’s used to pump out tons of code because upper management wants features released even faster than before. I’m sure the junior devs who don’t know a for loop from their ass will be able to learn and understand wtf Claude is shitting out
> AI is an incredible teaching tool. It's probably better for teaching/explaining than for writing code.
It is but how do you teach to people who think their new profession is being a "senior prompt engineer" (with 4 months of experience) and who believe that in 12 months there won't be any programmer left?
And how many will not?
For mist people it’s just a job to get money, they will put exactly as much effort in it as is necessary to produce an acceptable result
Only for people who wants be taught, this argument keeps coming up again and again but people in general doesn’t want to learn how to fish, they want the fish on a plate ready to eat, so that they can continue scrolling.
I see this a lot in juniors, they are solution seekers, not problem solvers, and AI makes this difference a lot worse.
I do agree it’s a great tool, so much better than trying to hope and pray someone on the internet can help you with “I don’t understand this line of code.”
I see so much creativity coming from young developers I just can’t agree. Yes most developers in the past 20 years who were only chasing big tech money were useless. Good riddance
Ai might bring forward the standardisation we never had. If coding dynamics shift enough then all the opinions about libraries and engines and frameworks might get less focused on readability and more on efficiency and easy composition by Ai.
Security gets outsourced to audited layers and Ai does the stupid boring jobs of gluing them together. Some developers become more specialised and niche, some pivot to product, some pivot to other areas.
There are plenty of people who joined software for the payout and hate it. Plenty of people who grown to hate it over time.
I've been enjoying using it to figure out toy projects but paying an API and depending on a service to code is very sour. I really hope hardware specialises and local models become good enough. Gate keeping development on centralised services would be a loss for everyone and ripe for dystopian outcomes.
I recently read a similar discussion in the context of AI in science and PhD students. And the point the author was making that the goal of having PhD students is NOT to produce academic research, but to train people. I think the same idea applies here.
Somebody still needs to train people, and the companies will probably need to ensure that they have resources for that, as there will not be enough senior people for all the tasks.
Every week, I read an article on the consequences of reliable coding agents in SWE industry. All such discussions on HN leads to a fundamental suspicion of the empirical scaling laws of LLMs or the infinite greed and short-sightedness of the market in inflating a bubble. I'm tired.
It's interesting to watch industry after industry hollow itself out from the inside then inevitably die long after all the financial people, investment bankers and management consultants have all cashed their checks.
Steve Jobs famously accurately called this out years ago [1].
Xerox, Boeing, PC manufacturers (who basically created the Taiwanese makers through a series of short-term outsourcing steps), etc. But there are two examples I want to talk about specifically.
First, one lasting impact of the 2008 GFC was that entry-level jobs disappeared. This devastated a generation of millenial college graduates who suddenly had a mountain of student loan debt (thanks to education costs outpacing inflation by a lot) but suddenly no jobs. It became a bit of a joke to poke fun at such people who had a ton of debt and worked as baristas but this was a shallow "analysis". It was really a systemic collapse. Those entry-level workers are your future senior workers and leaders. Those jobs have never come back.
The rise of DVR/TiVo and ultimately streaming brought on a golden age of TV in the 2000s. It was kind of the last hurrah for network shows that produced 22 episodes a year before streamers instead produced 8 episodes every 4 years.
But what made this system work was an ecosystem. Living in LA, Atlanta and a few other places was relatively cheap so aspiring actors and writers and entertainmnet professionals could get by with secon djobs and relatively low income. These became the future headline actors and senior professionals. Background work and odd jobs were sufficient. Background work also taught people how to be on a set.
Studios still had large writing staffs. Some writers would be on set. Those writers were your future producers and showrunners.
Part of what supported all of this was syndication. That is, networks produced shows and basic cable channels would pay to rerun them. Syndicating some shows was incredibly profitable in some cases (eg Seinfeld).
So the streamers came along and stripped things down. They got rid of junior positions. They adopted so-called "mini writing rooms". Those writers didn't tend to ever be on set. The runs were shorter and an 8 episode series couldn't support a writer in the same way a 22 episode series could. The streamers then were largely showing just their own content so residuals and syndication fees just went away.
All of this is short-term thinking. Hollywood has been both a massive industry and a source of American soft power internationally by spreading culture, basically.
I think the software engineering space is going through a similar transformation to what happened to the entertainment industry. A handful of people will do very well. AIs will destroy entry-level jobs and basically destroy that company and industry's future.
I predict in 10-20 years we'll see China totally dominating this space and a bunch of Linkedin "thought leaders" and politicians will be standing around scratchin their heads asking "what happened?"
I've seen a lot of hot takes on HN but faulting streamers for the inability of Hollywood to adapt to cultural and technological change is gonna require oven mitts. The rise of streaming is a win for the little guy and the viewers against Hollywood's capitalist entertainment oligopoly.
Yes and no. Often times managers are now asking ask Claude code to write it but I want it delivered tomorrow. This leads to us forcing to use LLM generated code without enough time to review or understand it.
"AI is making junior devs useless" is a dangerous and incorrect conclusion. If this idea is repeated too often, people may start to believe it and even quit studying computer science altogether.
First of all, developers who only learn to code in a short bootcamp are often not well prepared — but that was already true before AI. In the past, many junior developers were students who were learning programming while studying, not just people who took a quick Python course on Udemy.
Instead of declaring junior developers useless, we should raise the standard: learn how to code properly, how to maintain code, understand networks, and build strong foundations in math and computer science. A well-trained junior developer is still extremely valuable and will always be needed.
I assume junior devs can at least search. AI often doesn't even do that. That's why there are things like context7, which in a narrow context helps but not perfect.
There are lots of ambiguous situations where a search and human "inference" can solve that AI still can't.
I can tell the AI to do something, it uses the worst approach, I tell it a better way exists, it says it validated it does not, I link to a GitHub issue saying it can be done via workarounds and it still fails. It's worse for longer tasks where it always shortcuts to if it fails pick a "safe" approach (including not doing it).
Junior devs: you have an oracle you can pester incessantly. Make the most of it so you can learn to detect its mistakes, know when to push back, and what to ask of it. That's when you are in the clear. Juniors who merely parrot the LLM get fired.
You don't have to take 5 years off for it. Just continue same old business (assuming you don't have outside pressure to use this slop-machine) and keep your skills and judgment sharp until the tools and workflow stabilizes, and most of all, the money to fuel this hype runs out.
The reason I say that is because I don’t want to deal with trying to keep things from collapsing around me while people slam in tens of thousands of lines of vibe coded slop.
Companies will continue to demand it (I know people working at companies that are literally looking at AI usage as an individual performance metric, puke emoji), and probably 95% of humans using pretty understandable human logic aren’t going to work harder than they need to on purpose.
I wish I had a solution. I think the jury is still out on whether programming will be a dead profession in a short number of years, replaced by technical protect operators.
"juniors are useless": Maybe y'all should consider updating this hyberbolic language. Nobody is born a "senior developer", so surely all of your training as a "junior" is not useless. There is always a disconnect between what younger people know and what older people expect them to know, so training is required almost universally.
> If I’m reviewing your code and I ask you why you went with a certain approach, and you tell me “the AI suggested it”, I’ve immediately lost confidence in you.
I’ve experienced similar things and so understand the feeling, but this is poor leadership. If someone on your team makes it all the way to a code review and still thinks ‘the AI suggested it’, you failed to train them, failed to set expectations and they have justifiably lost more confidence in you than vice versa.
If we analyze the rest of the article through the lens of weak leadership, it sounds less like an AI problem and more like a corporate leadership problem.
The cost of generating code is now laughable, so that's not the economic value brought to the table by a junior engineer, or really, any engineer. The value is now generated by knowing what code is good code. You're going to have to have talks, book clubs, hackathons, and the like to get your juniors to know what good code is. Do they know what design patterns are? How about good architecture? If they can't name a few design patterns, you're not investing enough.
This is ridiculous. New developers will learn a completely different skill path from what we learned, and they will get where we are faster than we did.
When I started my career I heard people say almost verbatim "Stack overflow is making junior devs useless", with the idea all we did was copypaste scripts over. The same people failed, and the same people who can use the tools will succeed now.
You definitely did see a difference between people who just copy pasted from stack overflow, and from people with good fundamentals. The uncomfortable truth though, is that the industry didn't need good coders, it needed a bucket load of basic web apps and it needed bums in seats.
I think the irony of AI is going to be that it will make the remaining software jobs properly hard again, and implementers (ex coders) will be able to succeed with even less code knowledge than before.
I worked under people who started as juniors that way but were politically savvy. Or just ruthless. And pushed their way to the top by stealing projects, lying through their teeth, and other such tactics.
They were slowing down progress because their methods involved sabotaging the progress of others because it might make their own contributions shine a little less.
They were the cause of using libraries like leftPad all through business critical code, and cutting anyone down who dared to simply question why.
These things cause ripples. The smartest and most capable staff leaves, what results is a churn of the same kind.
But hey, they get a trip to Mexico every year and burn through millions every two years. Profit any day now.
I did my first completely vibe coded not looking at a line of code implementation last year and my second this year.
I could care less about why either Claude, Codex or before that a developer was using a for loop or a while loop. I did and do care about architecture.
I’m no more going to review every line of code with AI than I am when I was delegating to more junior developers. I’m going to ask Claude Code about how it implemented something where I know there is an efficient way vs naive way, find and test corner cases via manual and automated tests and do the same for functional and non functional requirements.
The "Junior Trap" is real: if you offload your thinking to Claude or GPT-4, you’re hitting "Done" for the day, but you’re accruing massive Learning Debt. You aren't building the failure-pattern recognition that actually makes an engineer valuable.
In a world where "Code is no longer a skill," the only way to survive is to stop being a "Prompt Operator" and start being a "System Auditor." If you can’t explain the trade-offs of the architectural pattern the AI just gave you, you aren't an engineer, you're just the person holding the screwdriver while the machine builds the house.
Nah, it makes teams useless. Maybe not quite yet, but soon, one engineer will be able to do a few sprint teams' worth of work, and deliver features orders of magnitude faster than a team working in parallel. Yeah, generally at first this will be seniors only. But before long, a junior will be able to come in and learn to manage one sprint team's worth of work under the guidance of a senior and partnered with a PM, and grow the product from there. Long term, I imagine 90% workforce reduction will be the norm. Just about all software is a rinse and repeat of some other software, not much true innovation, so picking and choosing and implementing some other software's feature into your own will start to become trivial single-day projects from start to finish. Hopefully AI creates some new industries that SWEs can roll into, but I'm feeling more doomer every day.
Yeap, and people are still forcing juniors to make small code changes when they should be learning by creating entire apps on their own, deploying them, etc. WITH a senior giving them feedback occasionally. I think people are going to take a while to catch on though, for better or worse....
Yeah, IMO one of the first things we'll see change is more of a migration back to monoliths. Right now adding a feature has to go through multiple teams, a dozen services, a coordinated implementation and deployment schedule, a Byzantine and often manual set of integration tests, etc. Yeah AI can help with that, but the point is that AI doesn't need it. On a monolith, it sucks for dev teams because parallel development at large scale is difficult and other teams' bugs can delay the launch of your unrelated project. Hence, microservices became popular. With AI, development happens so fast that it's largely serial. So there's no real coordination needed. A whole feature is one PR, one set of tests, one app to run locally if you want, one deployment, one thing to look at and roll back if there's a bug. Creates a virtuous cycle all the way up.
I imagine lots of established companies will struggle migrating back to that pattern, but I have to think most new companies will head in that direction, which should let them catch up quickly.
I maintain that in the future, any person wishing to learn any skill (not just coding!) will need to willingly eschew the use of AI when learning until they have "built the muscles". The literature is clear that repeated, hands-on practice is really the only way to build skills.
I suspect the progression will be "No AI until intuition (whatever that is for that skill)" -> "Gradual use of AI to understand where it falls short" -> "AI native expert".
How to actually implement this at scale is still TBD ;-) Ironically, AI will be invaluable for this e.g. as a hyper-personalized tutor but it will also present an irresistible temptation to offload the hands-on practice. We already have studies indicating the former is helpful but the latter stifles mastery. At this point I can only see self-discipline as a mechanism to willingly avoid AI.
Unfortunately, our testing-oriented education system only serves to incentivize over-reliance on AI (Goodhart's Law etc.) None of our current institutions and processes are suited for what is already happening and will only accelerate from here on. Things will need to change radically.
For this reason, I once predicted apprenticeships will be a thing again, and already there are signs with Microsoft's preceptorship proposal: https://dl.acm.org/doi/10.1145/3779312
This is highly encouraging because a tech giant is not only acknowledging the problem, but proposing a solution. Not a complete solution by far but at least a start.
Junior devs have always been useless. You used to give them tasks that take them a week or two even though a senior engineer could do it in a couple hours, not because you wanted them to contribute, but because you wanted them to learn to contribute.
The same ethos makes sense with AI, it's just that every company is trying to avoid paying that training tax. Why turn a junior into a senior yourself if you can get the competition to pay for it instead.
It's interesting that the same dynamic is playing out on a much larger scale with children. A child is far more helpless than a junior engineer - at least a junior engineer can feed themselves, wipe their own butt, avoid destroying the room, and generally keep themselves alive. Everybody wants to offload the cost of raising children to parents, because the economic benefits aren't realized for 25+ years yet the costs are very substantial (frequently, at least one parent's full-time attention, costing them an income). Prospective parents are saying "fuck that shit" and simply choosing not to have children.
The long-term effects are going to be much like the effect of the software industry turning away from juniors: total collapse. When you have no workforce, you'll do no work - hell, there is just...nothing, nonexistence, no consumers either. But the fertility bust operates on a longer timescale (I think the software industry will start feeling the dearth of juniors in ~5 years, the economy as a whole won't feel the dearth of children for ~5), and it's far more fundamental. Rather than one industry disappearing, all industries will disappear, likely refactored into something that looks far different.
It also reminds me of those ecological predator/prey/locust models that I studied in calculus class, where population dynamics for many species have a tendency to overshoot the carrying capacity of the environment. Each individual in the population makes their own reproductive & survival decisions, but the sum total of them leads to population collapse and a near total extinction, followed by recovery once the survivors find resources abundant again.
> When you have no workforce, you'll do no work - hell, there is just...nothing, nonexistence, no consumers either.
But for a beautiful moment in time, we created a lot of value for shareholders
https://www.insidehook.com/culture/story-tom-toro-new-yorker...
It's basically a "Tragedy of the Commons" situation across the board.
Sort of. With the salient point that nothing really exists, but the commons. The individual is nothing without the whole society around.
Also sort of. Some form of social organization seems to be necessary for humans to function. But humans are also pretty good (well - relatively speaking, it usually seems to require a war or revolution) at changing that form of social organization as technology, population, and environmental conditions change.
I think this is a very likely outcome. We aren't going to get continued population growth next generation; a significant number of the people needed for it will never be born. This is going to have ripple effects across wide swaths of political and economic organization. But you'll have pockets of population that basically barricade themselves off from the wider economic world and insulate themselves from its collapse, and then the people within them, along with whatever form of social organization they happen to adopt.
> Junior devs have always been useless
> The same ethos makes sense with AI, it's just that every company is trying to avoid paying that training tax.
Last time when a junior dev was added to my team I had a similar thought. But then talking with management I was informed that things went beyond just training.
The company had a social responsibility pledge and understanding with the local educational institutions. They had to pledge to be part of the internship and hiring activities every year. The company could not chose to be fair weather friends and try to recruit people only when they saw fit.
The other aspect was cost. A team made of only senior engineers was costly.
The last aspect was leveling up. Unless the company has lots of levels the team might end up lots of engineers at the same level. And with the inverted funnel nature of promotions it meant some engineers might end up waiting years for the promotion.
So, it was better to have teams with some junior, intermediate and experienced engineers. That way costs and promotion flows were controlled.
Now with AI the impact might go beyond junior devs. I see even the intermediate devs being impacted. It is more likely that companies think they can replace say 1 junior + 1 intermediate with 1 junior dev with AI. Or something along those lines.
Then don’t base comp on promotions - problem solved.
This kind of flippant approach is equally valid as "just use ai", "let other companies train the juniors", and "don't give promotions just hire new juniors + ai". All of these have obvious problems from their overly myopic viewpoint.
How is that flippant? I went from an “architect” over the entire cloud developmrnt strategy at a 60 person startup making $160K in 2020 to working at AWS ProServe making $220K (cash + signing bonus + RSUs) as an L5 (mid level - no longer there). Do you think I cared about my title?
I said just the opposite - hire fewer seniors + AI and don’t hird juniors.
I’m now a staff level employee again at a 3rd party consulting company working with small and medium businesses (my preference) after taking a detour in BigTech from architectural roles at smaller companies. My quarterly strategic goals were a combination of what the CxO’s/directors defined and what I defined was in the best interest of the company. None of those goals had anything to do with larger societal issues.
I’ve had some who are useful almost out of school. The amount of tickets is always growing and have someone pick up those “when things calm down I swear I’ll address this” tickets is always helpful. If they can’t get anything done by themselves in the codebase then it gets much harder. I do also think that some people have completely forgotten all the context they didn’t have when they started off xx years ago so mentoring is not always very good
What’s the importance of then learning to contribute if they will probably jump ship anyway when they get good enough? Your HR department is not going to give them a market rate raise to keep them - see salary compression and inversion. A junior developer just isn’t worth the investment.
I have never once told my manager “it would be really nice to have a few junior developers. It would really help us get this project done on time”. They do “negative work”.
Yes not having juniors become seniors is an industry problem. But my goal is to reach my company’s quarterly and anual goals - not what’s going to happen 10 years from now.
> I have never once told my manager “it would be really nice to have a few junior developers. It would really help us get this project done on time”. They do “negative work”.
I have. A good junior can do in a week what a senior with domain knowledge can do in a half day, with only an hour of mentoring along the way. This isn’t a great exchange rate per dollar (juniors are cheaper than seniors, but not that much cheaper) — but seniors with domain knowledge are a finite resource, you can’t get more of them for love or money, while juniors are fresh-minted every semester. The cheapest way to shipping may not go through juniors, but the fastest way usually does; and that’s completely ignoring the HUGE side benefit of building seniors “the hard way,” which is still easier than hiring.
And as a senior+ with domain knowledge, with AI I can do the work of two juniors without the communication overhead + do all of the project management, dealing with stakeholders, etc.
But you don’t build seniors, you build capable mid level ticket takers who jump for more money at the first opportunity.
And you can actually hand things off to them: this problem is now your problem. With AIs you’re herding cats
How is that true? As an architect when I was working at product companies, the director/CxO wasn’t going to call the junior developer to the carpet for not getting a task done or even the mid level ticket taker. Hell they don’t even know what the individual tasks are. I’m going to be the one ultimately responsible either way for project success.
Even when I was working at AWS as a mid level/L5 in the Professional Services division (lower title/lot more money), I was the one who was responsible for my “workstream” on larger projects. I couldn’t say “that’s not my fault. Blame the new L4 junior consultant who just got out of the internal boot camp for new grads”.
Now that I have moved back up to a more senior position in consulting, if a project I’m leading goes sideways, I can’t tell the customer that it’s not my fault, it’s the fault of the workstream leads and the workstream leads can’t tell me, it’s the fault of the more junior consultants who work under them. They will never talk to the client. The people under the workstream leads may not even speak English.
And that’s not meant to be an insult. The workstream leads have to he able to speak passable English and they can work with people under then who only speak Spanish.
Treat your employees well and they won't jump ship.
Its not about treating them well. The environment and culture isnt the problem. Its that tech companies have decided that they dont like giving appropriate raises to current employees. They are somehow fine with paying the same or more money to bring in an equally qualified external hire. But not with retaining people. Idk why, maybe some stupid MBA rule, mayne theres some good reason behind it
The problem is that (hypothetical) you as a line level manager don’t control comp and raises. Even in BigTech your manager doesn’t control your promotion and everyone knows it’s better to “boomerang” because you will get paid less being promoted to an L5 (mid) from an L4 than someone hired as an L5.
> But my goal is to reach my company’s quarterly and anual goals - not what’s going to happen 10 years from now.
Your benefiting from the work of peopke who did worry about what will happen 10, or even 20 30 years down the line. People like you are why the rides gonna stop
So I should become a director or CxO of a company (because line level managers are powerless to do anything) so I can make those types of decisions?
Not that I’m a line level manager - I’m just a high level IC who is at the same position on the org chart as a line level manager
It’s telling that OP worked at Amazon because this pretty much sums up the culture there - new grads churning through and burning out in two years while a few seniors stick around and perpetuate the cycle.
I was 46 when I was hired and it was my 8th job out of college. I was far from a naive college grad. I went in with a purpose - keep my head down for 4 years through my initial four year package, put it on my resume, build a network and leave for a smaller consulting shop.
In fact the only reason I got my job there was because a recruiter reached out to me about an SDE job on the retail side. There was no way in hell that I was going to sell my big house in the burbs of Atlanta (at the time), uproot my life after Covid and work for Amazon. I knew what I would be getting into.
She then told me about a “permanently remote”,”field by design” role at AWS ProServe. I was like sure why not?
>What’s the importance of then learning to contribute if they will probably jump ship anyway when they get good enough? Your HR department is not going to give them a market rate raise to keep them - see salary compression and inversion.
Obviously that hasn't historically been true, else there wouldn't be any senior developers as companies would have wised up to that and nobody would hire them as juniors.
- Not everybody is a job hopper (even in Silicon Valley one sees that most junior FAANG devs stick around for a good while).
- The HR department is absolutely going to give junior developers that pass the cut after a year or so a market rate raise.
- In limited hiring periods, they'd be grateful to have the chance to stick around, while in bullish "boom" periods companies can afford to spend to keep people, expand and give them bigger roles, and so on. It's in the in-between that it becomes more problematic, but now we're in a "limited hiring" era.
>Yes not having juniors become seniors is an industry problem. But my goal is to reach my company’s quarterly and anual goals - not what’s going to happen 10 years from now.
That's how companies fail.
It's also not a good strategy at the personal level. If you command more devs, you get more leverage.
This is not true - the average tenure for a developer across the industry has been 3 years for well over a decade.
> The HR department is absolutely going to give junior developers that pass the cut after a year or so a market rate raise.
This is also not true from small companies to FAANG - see “salary compression and inversion”
> That's how companies fail.
The company failing in the long term is really not any current employees main concern unless you are a founder if the average tenure is 3-5 years. Even the stock market doesn’t care about the long term viability of a company.
BigTech for instance can afford dead weight. Amazon has an internship program and for those who come back or through their non traditional programs for their internal consulting division (AWS Professional Services) they have a 3 (6?) month training program.
In ProServe at least (former employee) even for their l5/L6 employees, they have the 3 month training program - “AWSome Builder” where you simulate a customer project and have to pass.
After leaving AWS and being hired as a staff consultant by a third party company, they put me on a plane two weeks in to meet with a customer. They don’t even hire less than senior+ people in the US.
If the median tenure is 3 years and the software business still is very profitable then people must be net useful within that 3 years. A lot of people also just don’t want to job hop much and honestly the interview culture keeps me from hopping more. I do still fall in the 3 years per hop but I’ve always had a good reason to- ie. layoffs, company going in a worse position than when I started, shit management at various levels, forever compounding responsibility…
>This is not true - the average tenure for a developer across the industry has been 3 years for well over a decade.
That counts temps, people who weren't a good fit and were let go early after hiring, mass layoffs, and mixes mixes startups and FAANG and consulting churn, none of which is the typical corporate IT worker scenario, and all of which bring the average down (but are not "hopping").
Corporate IT, government IT, smaller SMEs, and stable SaaS, have higher averages.
Nope, it doesn’t count temps at all. This is easily Googleable.
It doesn’t make sense to hire juniors at all other than as a service to society. I haven’t hired a junior in 4 years. The one I hired 4 years ago was because not only did he do reasonably well on the interview but he literally begged me because he trained himself to do it while painting houses so I saw a lot of passion in him.
Completely misses the point of training someone
So exactly what is the point for a profit seeking company to train someone except thsr you expect them to bring you more business value than they cost during their tenure?
This is the difference between being an engineer and being a clock puncher. You don't care about the business, you don't care about the product, you don't care about society as a whole. So long as you get your paycheck and your annual pay bump, fuck absolutely everyone and everything else, right?
Don't worry, just leave all your problems for someone else to fix. I'm sure that won't have any lasting consequences at all.
I work for one reason and one reason alone - to trade 40 hours of labor for money to support my addiction to food and shelter. The company is not going to give me money for “caring about society”. They are going to give me money to meet my quarterly goals to help them meet their profit goals for the company and in a former life , to make them look good for the public market pre-IPO and at another company for an acquisition.
I give a company 40 hours a week and all of my 30 years of industry experience and they give me money (and in a former life RSUs)
Sweet summer child. I was once opinionated and driven as you are now. I remember when I got out of college, I also thought like that of the mediocre clock punchers.
Now at my 45 years, I couldn't care less for whatever grand objective the current company I work for has. I exchange my knowledge and time for hard cash, and let the owners , ceo and whatnot run with their grandiose vision.
I only want to be left alone.
We all get here. It's funny when we turn back.
Funny enough, the more I got into this mindset, I slept better, made more money, and got more autonomy.
Once directors and CxOs know that you are completely aligned with the business goals and ignore everything else that “doesn’t make the beer taste better”, they trust your judgement and basically leave you alone.
Welcome to capitalism. Hire seniors and pay them 400k
Until there are no more "seniors"...
More money for me until I retire - well actually I’m past the point where I chase more money - and then after that - it’s not my monkey and not my circus.
Again HN bubble thinking. Most developers in the US are working at banks, airlines, insurance companies, etc in second tier cities - in the “enterprise” and are not making “$400K”. Most developers will never in their career see more than $175K inflation adjusted and I really haven’t seen comp on the top end increase in nominal terms in a decade [1] for enterprise devs.
That leads to my second point, in second tier cities, you see comp go from around $80K —> $115K -> $150K —> $175K, junior -> mid (pull well defined tickets off a board) -> Senior (leads larger initiatives) -> Senior+.
For instance look at what Delta airlines pays based on Atlanta.
https://www.levels.fyi/companies/delta-air-lines/salaries
Why hire a junior at $80K when you can poach a former junior now mid level ticket taker for $115K?
[1] after pivoting slightly to cloud + app dev customer facing/hands on keyboard consulting, I’m at a new plateau that’s higher.
Agreed. We are still in a capital crunch so overhiring is out of fashion. People don’t remember the early 90s or the dot.bust when the same things were said.
Kraft 1977 Programmers and Managers talked about this if I recall. Still the best alternate take on our industry I have ever read.
Yep. This is why many companies have a terminal level with “up or out “ rules. Before that level you are not fully independent and require too much supervision. No one wants a Jr engineer with 10 years of experience.
I see a lot of Sr engineers get very frustrated by how much time they have to spend helping Jr engineers. But, that’s the job, or at least a big part of it.
Or at least it was.
I burnt out helping a junior on my team for the past few months. It was just terribly obvious she was feeding my responses directly into a chatbot to fix instead of actually understanding the issue. I can’t really even blame her, there isn’t much incentive to actually learn
I've been in situations like that. For me, it's like interviewing, I just keep backing off, lowering the bar, making it easier and easier until they can get it, then start going back up again. I pretty quickly get a confident read on where they are.
If at that point it's clear (to me) the situation is not salvageable, it's a management issue, I've done my job.
That sounds like a bad hire, not a junior. Why didn’t your manager help fix that?
I gather that quite a lot of companies are using dumb metrics which would show this as _good_ behaviour, these days.
Sure, but the overriding metric should be the opinion of the Sr engineer who is supposed to be mentoring and supervising the Jr
You'd hope, but, y'know, AI mania.
Strongly disagree with this. Bad junior devs might be useless, but I’ve seen good ones absolutely tear through features. Junior devs fresh out of school typically have tons of energy, haven’t been burned out, and are serious about wanting to get work done.
And how do they compare to what a senior dev can do with Claude Code/Codex?
I bet you a senior could do with one good prompt to Claude what a junior would take a day to do before AI - and take time away from the senior.
Pretty favorably, because the coding agents suck.
> I bet you a senior could do with one good prompt to Claude what a junior would take a day to do before AI
It would still be a waste of a seniors time to write that prompt. They should have more important things to spend time on
And it’s not a waste of their time to have to give detailed requirements and troubleshooting steps to a junior developer, constantly being interrupted, and then having to check their work thoroughly?
If you have to be that detailed anyway - you might as well use AI.
No, teaching the next generation of humans is not a waste of time
I'm very sorry for you that you think that way
So exactly how am I going to convince my management to open a req for a junior developer who is not going to help us meet our quarterly goals and take time away from the other senior developers that will either have to work longer hours or do less work?
I’m not going to work as a charity and neither are any of my coworkers. We are all here to exchange labor for money.
We as a collective need to convince our management of this, but that needs to start with people getting their heads out of their asses and working together instead of this mercenary attitude you have
I don’t have to do anything except keep my head down, do my job and enjoy my well earned autonomy. I’m definitely not going to try to convince my skip, skip, skip manager to change their hiring policies. It’s not like my line level manager has any power over anything
Even when I was at a startup before 2020 and I did have the ear of the CTO and the founders I knew my ultimate mission was to do what was needed to get acquired and before that I knew exactly what my mission was when k was hired to lead the tech initiatives as we were acquiring companies “find efficiencies” and go public.
Or do you think I could have convince anyone of anything as an L5 at AWS in the middle between architect at a startup and my current company?
I’ve gotten plenty of use out of junior devs. The critical bit is what makes anyone a useful worker. I’ve found anyone that’s both dedicated and meticulous is worth the investment.
Sure there’s a wide range of skills and you can’t just hand any task to anyone and expect it to work out but some fresh collage graduates are more capable than the average person with 5 years of professional experience. At the other end you need to focus on whatever they actually are capable of doing. 40+ hours a week can slowly expand even an extremely narrow skillet as long as they’re a hard worker.
And have you compared the output of three junior devs to hiring one mid level ticket taker who isn’t that much more expensive + AI coding agents?
You don’t want 3 Jr devs at the same time because of diminishing returns. Most projects have grunt work where attention to detail is important but experience doesn’t really help much. AI can quickly come up with alt text for images, but ensuring it’s actually useful for someone using a screen reader is a different story.
1 new Jr every 2 years works quite well for a team of 7+ developers.
And that junior dev will probably leave in 2 years because HR won’t allow you to give them a raise to match their market comp.
But with a team of 7, I can’t believe that giving them all a Claude Code subscription wouldn’t be much cheaper and much more productive than hiring a junior. A junior though with AI is more dangerous than a junior without.
That’s not really an either or situation, you can have 7 team members with Claud Code and 1 Jr without.
I’ve seen many people stick around for a surprising amount of time while severely under compensated. I suspect that’s why HR does it.
What’s the use of the junior? You have just widen the gap between mid and junior developers making them less useful
Working on problems that are a poor fit for a mid level developer with AI assistance of which there are still plenty.
I find AI tools make everyone better at exactly the same kind of problems, which means a larger percentage of what’s left over is more cheaply done by Jr’s working without AI assistance.
A mid level developer is just a ticket taker. That’s their job. No matter what their title, if they are just pulling well defined tickets off the board, that’s what they are.
> I find AI tools make everyone better at exactly the same kind of problems
Fascinating. I am finding the exact opposite, like complete 180
How exactly do some people leverage AI to rearrange the furniture in the conference room for an office party?
Physical labor may be a tiny fraction of what a team of developers do, but I’ve seen what amounts to several thousand dollars spent on that kind of silly task because teams leverage the tools they have.
You pay contractors and caterers - that’s a poor excuse ti have dead weight.
You just replaced one task that an AI is useless for with a different task that an AI is useless for.
The point obviously still stands, and no I am not suggesting using Jr devs for physical labor alone is worth adding them to the team. Rather that “Work” includes a very wide variety of tasks that need to get done.
> I find AI tools make everyone better at exactly the same kind of problems
this part I find fascinating. most places I worked (30 years in, 10 as a contractor so quite a lot) the distinction between Jr, Mid and Sr falls exactly into the kind of works that they do. I, as a Senior, often work on hard shit, Jrs are not (yet) entrusted with those problems and work on entirely different set of problems. I cannot compute how AI makes everyone better at exactly the same kind of problems, problems I am solving today few people on my team are working on and then same goes down the "pyramid"
> Junior devs have always been useless. You used to give them tasks [...] not because you wanted them to contribute, but because you wanted them to learn to contribute.
Junior devs are by your own explanation not useless. They are the most important human investment in your project.
I mean, if we’re doing this, let’s be honest and go as far as mid-level engineers whose work needs constant correction, as well as the many, many senior engineers out there who are senior only because they lucked out in getting the title during the artificial dev scarcity of the ZIRP eras.
I think the idea of 'junior' needs to be refined a bit. By the time I got my first job I've been coding for years, and have built rather substantial things. In fact, in terms of pure coding ability, I was probably past the initial, fast part of my growth.
As should have others, which the university education system should have made sure.
The fact that some people come out of 4+ years of software engineering education utterly clueless means that they somehow managed to dodge having to build anything, I think means that they will never get good at any point in time, as they either were very talented at dodging having to build things, and I don't think that talent is going to abandon them, or they couldn't really grasp the basics in an environment designed for just that.
With that said, I think you can see for most juniors, what you can expect out of them in terms of pure coding ability - sure a lot of them have room to grow, but I've met so many great people who were very young, yet were useful from day one.
In fact, if you have the willingness to grind away at some problem, that puts you ahead of a significant amount of the pack. I have had the misfortune of working with people who lacked any demonstrable skill, and had coping strategies for having to deal with any sort of hardship. Getting useful work out of them was a challenge in of itself.
These people managed to get the years in to be considered senior, and are probably dispensing their wisdom 'mentoring' juniors somewhere else, and are no longer expected to actually contribute to meaningful issues.
I'm not sure if enthusiasts are the exception rather than the norm? I've noticed in the last few years, a lot of junior engineers do not have much active coding experience outside of their university education, they aren't the traditional "obsessed with computers and programming as kids".
There has been a much higher demand for software developers over the past 10-15 years than there are people who are obsessed with computers and programming
If you look at the general topic shift on HN over the years it's obvious most people are getting into tech because they want power and money, not for love of tech
> If you look at the general topic shift on HN over the years it's obvious most people are getting into tech because they want power and money, not for love of tech
I think that is a bit presumptuous, even if there is a hint of truth there. Just because a kid didn't grow up around computers and programming shouldn't exclude them from the tech path or mark them as imposters/pretenders.
Why? Because I learn every time I do.
Many juniors are actually very experienced but the industry can’t see that on paper.
> ...not because you wanted them to contribute, but because you wanted them to learn to contribute.
Rather because you want them to go away, because management conveniently forgot to reduce your load to account for time spent on mentoring.
> Junior devs have always been useless. You used to give them tasks that take them a week or two even though a senior engineer could do it in a couple hours
We havent dont it and I never seen something like that.
everyone was junior at one point, everyone. “junior” is just age mate, just age…
> > For someone like you, who likely has years of experience without LLMs, your brain totally understands good code/bad code, good architecture, and just general intuition around code and systems. LLMs must be an absolute gamechanger. But for someone like me who is starting out in this field, how am I supposed to build the years of experience and intuition that comes from manually writing code and building systems when companies are expecting AI to be used from here on out?
Answer: juniors need to work with seniors, and the seniors need to teach the juniors, and the juniors need to learn to use LLMs to learn, not just to do the work.
Next year I'm going to experience my first junior, which is crazy considering I have 3 YOE. Nonetheless I'm very curious what kind of graduates come out of the current education
As I tell my students: juniors, you must write the code
https://htmx.org/essays/yes-and/
Everyone else: we must let the juniors write the code.
Seniors come from juniors. If you want seniors, you must let the juniors write the code.
> Seniors come from juniors. If you want seniors, you must let the juniors write the code
The average tenure of a person in engineering role is so short that very few employers are thinking about developing individuals anymore.
The actual way this gets approached is "If you want seniors, you must hire seniors".
I'm not sure how this plays out now. But it's easy to imagine a scenario like the COBOL writers of the last generation.
It's a self inflicted wound. Companies do not reward loyalty. They do not give out raises congruent with what you can find if you leave. Business-types unirionically think seasonal layoffs is a "good thing." Self hemorrhaging your institutional knowledge is insanity
Is it? Seems to be working fine for most
The issue stems from 2 things:
1) People hearing "an LLM is as smart as a junior" and actually opting for the LLM subscription price instead of hiring a junior
2) The gap between senior and junior in terms of performance has become larger, since the senior devs had their hands get dirty for years typing stuff out manually AND also tackling challenges.
This generation of junior-mid developers will have a significant portion of the "typing stuff" chopped off, and we're still pretending that this will end up being fine.
I think your second point is interesting, and it has actually already happened a couple of times.
It used to be a lot easier to find devs that knew assembly and could navigate call stacks through memory by hand because a lot of folks had to learn that to get their job done. Now higher level languages have mostly eliminated that level of operation.
The same applies to infosec roles. It is 10x harder for junior infosec folks than 20 years ago because there are a bunch of skills you need in infosec that today's mainline dev experience doesn't need, but were more common a while ago.
Case in point, I remember working with a partner company's junior engineer on some integration. They needed some hard-coded constant changed and time was of the essence. I told them to change a couple bytes in the elf binary directly. They looked at me like I was a wizard. I thought it was a fairly pedestrian skill having grown up reversing computer game save files.
Hex editors are a black box to most 99% devs these days. I noticed their use falling off once code-signing came into use.
The challenge is to get cost sensitive businesses to support this. Juniors are a cost and when trained move on, thats the fundamental problem. Retention only works with smart companues, for most other companies its a revolving door.
On the plus side, as a dev with 30+ years of experience, I am commanding a very good contract salary these days. Revolving door companies stuck in process hell and product rot, and cannot deliver new value, so they’re scrambling to find experienced devs that cost a premium. My salary today makes up for peanuts at the start of my career.
The real question will be; Do we need to pay the juniors to write code to become seniors?
If coding is an art then all the juniors will end up in the same places as other struggling artists and only the breakout artists will land paying coding gigs.
I'm sitting here on a weekend coding a passion project for no pay so I have to wonder.
So non technical business people will hire vibe coded seniors?
> Seniors come from juniors. If you want seniors, you must let the juniors write the code.
Companies know this as well, but this is a prisoner dilemma type situation for them. A company can skip out on juniors, and instead offer to pay seniors a bit better to poach them from other companies, saving money. If everyone starts doing this, everyone obviously loses - there just won't be enough new seniors to satisfy demand. Avoiding this requires that most companies play by the rules so to say, not something that's easily achieved.
And the higher the cost of training juniors relative to their economic output, the greater the incentive to break the rules becomes.
One alternative might just be more strict non-competes and the like, to make it harder for employees to switch companies in the first place. But this is legally challenging and obviously not a great thing for employees in general.
The way other professions do this is by burying trainees with debt and then writing off debt if they stay.
See Revature.
Not every career path starts at a software first company. Not every software first company works on the most intense codebase.
And therefore in my experience not every senior engineer would hack it as a senior engineer at a more intense company myself included.
This isn’t a software unique experience. It’s life.
It’s already getting harder to find juniors willing to write the code and harder to discern whether someone is as willing as they say. And I feel like asking junior to make this decision and just have self control is a tricky double edged sword. Even if I want them to (and I do!) the competitive and ambitious juniors I suspect will still lean into AI code gen heavily as it makes them look better and seem more productive. Seniors probably need to do more than let them write the code, we probably need to figure out ways to encourage, require, or even enforce it at some level, if we want it to happen.
before you had a lesson that every engineer has to start with writing C, yet most of modern devs never did.
Seniors should be prepared that Seniority will mean different thing and path of getting there will be different too.
Just like there was a shift from lower lvl languages to high level
I agree with the sentiments here. But, I’m less hopeful about the presented solutions.
I think my argument against humans still needing to know how to manage complexity, is that the models will become increasingly able to manage that complexity themselves.
The only thing that backs up that argument is the rate of progress the models have made in the last 3 years (ChatGPT turned 3 just 3 months ago)
I think software people as a whole need to see that the capabilities won’t stop here, they’re going to keep growing. If you can describe it, an LLM will eventually be able to do it.
disagree because when the "super fast" new CPUs of 20 years ago became common, it was easy to write code that executed slower than previous code, due to language constructs and wasteful work patterns. Therefore, I predict that LLM code can explode in complexity (14KLOC for a binary file parser with some features) but that compute will bog down and effort to understand will explode.. that is, in extreme cases.
> If you want seniors, you must let the juniors write the code.
I do not want more juniors, because given time they will be my competition.
Ok but even pre ai I felt like each years interns wanted to take as many shortcuts as possible and not learn.
I think the allure of high TC (150k base or more for entry level) led to many non engineer brained people to enter tech.
Many people can do rote memorization, it’s even ingrained heavily in some cultures iykyk. However they can’t come up with much original or out of the box thinking.
The silver line is not to avoid AI but to use it wisely, and this is coming from somebody who used to hate AI.
I am not a Python developer or developer for the matter but Perplexity AI did help me to understand the bsic of Python for API requests and get projects delivery with 94% code coverage and vulnerability free.
AI also reduced the time spent with Ansible playbook generation, but I do know Ansible, I do know Linux, homelab is my hobby so I am not just doing copy and paste. I review whatever it generates and correct it when required.
In the companies I have worked and work, I see developers themselves confessing "I used AI, it works but idk how"!
AI itself does not make you useless, only and only if you used it as a smart search engine.
If you are doing copy/paste, you are going to get so screwed professionally speaking.
Folks are no longer learning and what they are doing, AI can do on its own.
That is making some developer useless.
My nightmare scenario (which might start to materilize) is that our last years in the industry will be becoming prompt monkies / agent "managers" working on codebases we barely understand in such velocity there's no way we can gain real understanding. Whenever something breaks (and it will , a lot) A.I will fix it - or so we'll hope. And the sad thing is - this might work; you'll get more stuff done with fewer people. Sure, we didn't sign up for this, it's not a fun job what I've described, but why should management care? They have their own problems and A.I is threatening their jobs as well.
At work we build enterprise software with stuff like Kotlin+Spring + multiple NextJS apps + Microservices + Rust CAD engine.
I haven’t have written code aside from tweaking stuff here and there in probably 3 or 4 months. Before that I wrote code by hand every day for many years.
I’ve found a lot of fun parts of my new workflow that I enjoy. I still miss being fully immersed in a problem deep in the files… and sometimes it feels like homework reading so many implementation summaries from Claude because the feature spans 4 repos and is too much code to read. But I do love shaping the code into different solutions exploring in a way that is unique to ai native workflows. And I love building agent skills and frameworks with/around them and expanding it out to more aspects of the company or life — there’s deep work to be had that still feels like hacking in the trenches. I get a lot of the same satisfaction in different ways, and there’s a lot of exciting novelty to explore that was previously out of reach due to time and energy constraints.
Also I don’t like our backend stack and I hate React / NextJS to the degree of derangement syndrome — I am so happy that I don’t have to write it and I can just focus on UX, making customers happy / lives easier / shaping the software into better and better versions of itself at such a faster pace.
People who learned good software engineering intimately before the inflection point are extremely lucky right now. Existential dread and the stages of grief have been a part of the journey for me too sadly, but there’s a lot to celebrate and explore with the right attitude.
Similarly, I started using Claude to add some features to an native app on iOS and Android in early January. That was so successful (tooks days instead of weeks) I started applying it to client work and basically haven't written any substantial code since. A big change from around 40 years of writing code pretty much every day and I'm enjoying the increased velocity from not having to web search for API and CSS syntax details.
My son, working in another dev company, reports the same - he basically hasn't written much actual code for about three months. It's a massive change.
I feel the same way, I have many years of experience, and I have gone from writing everything by hand to using claude code all the time (my latest company is very pro doing everything with AI).
Since I have been a software architect for the past 7-8 years it feels in some ways that that experience makes using claude code a lot more productive than for my non-architect colleagues, as I am able to steer it much more effectively whether directly in sessions or via custom skills / mcp.
The big issues right now for me are hiring and manager expectations, I changed positions last fall due to mass layoffs and it took me 3 months to find one: having leetcode interviews in the current climate seems completely useless, even more than it was in the past, and system design interviews are so formulaic it also feels like a crapshoot. Plus every job getting hundreds of AI generated applications makes actually being considered in the first place quite difficult.
Manager expectations are also ridiculously inflated nowadays, it seems most action items that come are claude written with fantastical random statistics (if you add caching you can make your backend 98.3% faster!), and it takes so much time to fight this and unrealistic team velocity expectations.
Interesting times, I do feel lucky I have had a long career, but I very much fear the ladder being pulled up even more than it has been when outsourcing because widespread. I know everybody says "things always change, new opportunities will open up to compensate for the ones that are being lost" but this time it does feel different, and not in a good way.
Things are changing so fast and so chaotically with this technology. I'm also writing everything now using Claude code, and I've been thinking a lot about what this means for my work moving forward. One thing I've noticed, is that I will just keep hammering and hammering on my work until I force myself to quit. Even on the weekend I feel the pull to go work on it. I'm just less sort of mentally exhausted by work, I suppose, but I don't think that's particularly healthy if it leads to me working way more than I should. On one hand, I think that's a reflection of how powerful and exciting this technology is, but on the other hand, I think that it triggers some different kind of reward function in my mind that I'm not used to.
In any case, I think if one wants to continue to have a career in this industry for years to come, it's basically table stakes to become fluent in using these tools.
It already happened. The old timers correctly observe that modern applications are bloated and inefficient because of all the heavyweight frameworks, excessive abstraction layers, and "left-pad culture" where external dependencies are pulled in to do the most trivial things but that these things enabled less capable developers to effectively build software to fulfill industry demand. LLM-only coders are just the next step in the devolution.
That's literally what https://news.ycombinator.com/item?id=47157039 seems to be about, and I had the same reaction as you
> My nightmare scenario (which might start to materilize) is that our last years in the industry will be becoming prompt monkies / agent "managers" working on codebases we barely understand in such velocity there's no way we can gain real understanding.
It will always be preferable to work on an understandable codebase, because that maximizes the AI's affordances too. And then the AI can explain things to you. A skilled human will always have a lot of solid knowledge relating to their hyper-specific niche that isn't part of your average general purpose AI, so humans will obviously have a key role to play still.
I'm already seeing this in the company I recently joined: 80-90% of code is generated/prompted. Big PRs, very little review or oversight. Absolutely nobody considering long-term architecture (and IMO nobody capable of such). In general, there's very little critical thinking involved at any stage, just throw error messages back into the LLM, rinse and repeat. I'm hoping there's a world where people with skills are useful in getting these projects back on track, but perhaps as a society we're learning to accept this reduction in quality.
Your codebase is essentially becoming bytecode
And how do u sum up the tradeoffs so far, or is it too early to tell ? Do u see lots of unacceptable shit making into production that wouldn't have before A.I for example ?
This is 100% an issue on the side of the senior developers. Imagine saying "these juniors are useless" because you are making them work in assembly, but C has just been released. You are giving them menial work that is no longer required to do by humans. Instead of giving them the task "update these email templates", the norm should be: "create this new service that automates an internal process". They will make mistakes and they will learn - but what they will be doing is going to be very useful and also give them chance to grow the necessary skills for this new era, with the supervision of a senior.
I think the issue is they used to make progress at a snails pace and you had plenty of teachable moments.
Now anybody can vibe code something that seems to work with a million landmines.
How can they develop the intuition around this given that they don't know what they don't know? How can we review it and help them get there?
Maybe we can figure it out, but I'm not sure it's easy or obvious.
By making them walk into the landmines and forcing them to fix it - that is how everyone became a good programmer. It's just the scope that has changed.
Actually the truth is that a lot of senior devs are not very good either, and have negative value. But they have an inflated value of themselves that does not reflect reality.
Pretty much all software projects seem to peak, and then decline in quality. There are only a handful of senior devs in the world who are actually good programmers.
I agree. But it’s not about that they have inflated value but it comes down to how “modern” software producing organizations work. Product managers and C-level people do not know what they are doing either. Most people are part of the “software engineering” theater, recruiters are recruiting, manager are managing, software developers are developing software, all of them just to get paid or gain status in the org. Most of the values come from that handful people who can really deliver.
Yeah read Software and Mind by Andrei Sorin
> The mechanistic myth is the belief that everything can be described as a neat hierarchical structure of things within things. And few of us realize that our entire culture is based on this fallacy. While the world consists of complex, interacting structures, we prefer to treat every phenomenon as a simple, isolated structure.
> Through our software pursuits, the mechanistic myth has spread beyond its academic origins, and is now affecting every aspect of human existence. In just one generation, it has expanded from worthless theories of mind and society (behaviour- ism, structuralism, universal grammar, etc.) to worthless concepts in the field of programming (structured programming, object-oriented pro- gramming, the relational database model, etc.) to worthless software-related activities that we all have to perform.
> What is worse, our mechanistic beliefs have permitted powerful software elites to arise. While appearing to help us enjoy the benefits of software, the elites are in fact preventing us from creating and using software effectively. By invoking mechanistic software principles, they are fostering ignorance in software-related matters and inducing dependence on their systems.
> Increasingly, in one occupation after another, all we need to know is how to operate some software systems that are based on mechanistic principles. But our minds are capable of non- mechanistic knowledge. So, when the elites force us to depend on their software, they exploit us in two ways: by preventing us from creating better, non-mechanistic software; and by preventing us from using the superior, non-mechanistic capabilities of our minds.
> The ultimate consequence of our mechanistic culture, then, is the degradation of minds. If we restrict ourselves to mechanistic performance, our non-mechanistic capabilities remain un- developed. The world is becoming more and more complex, yet we see only its simple, mechanistic aspects. So we cope perhaps with the mechanistic problems, but the complex, non-mechanistic ones remain unsolved, and may eventually destroy us.
Wow, I'm immediately hooked. The book is free online as well to download.
This post, ironically, seems very likely to have been written by an LLM :/
"it's not x, but y", with bonus em-dash:
> your value as a developer is not in your ability to ship code. It’s in your ability to look at code
"But here’s the thing."
"And honestly?"
I think so. Just go to the homepage, all of the thumbnails are AI generated with clickbaity titles.
It feels like to me that junior devs don’t understand what they even need to learn. They just use agentic coding to get things done, without any deeper knowledge.
The worst is, they think they know exactly what they need to learn, and also think they can make good decisions.
I can't seem to get the article to load, but I think I get the gist from the title.
I hired a junior "dev" who literally hadn't even completed an HTML course. Before AI I could not have hired them because they literally did not know how to dev. After AI, anyone with a little grit can push themselves into the field pretty easily.
As with everything in life: you can choose to hard route or you can choose the easy route and your results will follow accordingly.
Adam, can you please share, how in the world, this junior dev got hired with you?
I'm self-taught dev with multiple years of experience. I choose the hard route, even after AI. For me, programming is theory building, so I always choose understanding above all else.
Rock solid understanding of TypeScript, frontend and backed.
I have sent 100s of CVs. For Juniors, Mids and Seniors. Not even a single interview.
I will be glad for your thoughts on the matter.
> As with everything in life: you can choose to hard route or you can choose the easy route and your results will follow accordingly.
Hard agree, but probably not in the way you're implying.
It's the difficult things that make life fun and interesting. A life spent going from one easy thing to another is a life barely lived at all.
From a professional standpoint, I’ve never found “enjoyment” from coding. I enjoy every part of the process of getting business goals from talking to stakeholders -> completed project with coding being the necessary evil.
Funny enough, I started working in 1996 professionally (and had been a hobbyist for six years before going to college). But it was only between 2012-2016 that I was a ticket taker without working with the end user directly - everything I’ve done has been B2B.
GenAI (and working remotely since 2020) has made me enjoy every part of my job.
> I hired a junior "dev" who literally hadn't even completed an HTML course.
I mean, I'm a fairly senior dev, and have literally never completed, or indeed really heard of, a html course. Is that, eh, part of your average CS degree these days?
25 years doing distributed systems and best i can offer anyone is
<marquee><h1><blink> welcome to cyberpunk’s j33t website!!!
:) :)
Fun thing about that:
> Netscape has stipulated removal of the <marquee> element from the Internet Explorer during an HTML ERB meeting in February 1996, as a condition to removing the <blink> element from the Netscape
It's like nuclear disarmament treaties, but for annoying things.
Damn, my elite frontend skills are 30 years out of date?
Better stick to apis then I guess.
>> who literally hadn't even completed an HTML course.
so what is their value? proxy your requests to ai?
Why are you paying them instead of running the AI yourself?
I can't wait until the AI people realize that without developers' original ideas, AI has nothing new to steal from. We don't create, AI will spit out the same old concepts. What, you're gonna create the next generation of AI by training it on what the very same AI has already produced? C'mon now.
You don't get technical creativity reflexes by using AI. This is technical stagnation in the making. By cannibalizing its own sources, AI is ensuring that future generations are locked-in subscription models to do the most basic technical tasks. This is all obvious, yet we speed up every chance we get.
It might be a mistake to assume tomorrow’s training looks like today’s. Unsupervised learning is a thing and a very hot research topic, precisely because it avoids some of today’s big problems with acquiring the vast amounts of training data necessary.
Unsupervised leanring has been around for years and is already how the current wave of models are trained. It doesn't mean no data, it means no human provided labels of the data. So you still need creative new human ideas to move LLMs forward. LLMs != intelligence.
Exactly this. Even frontier models like Opus 4.6 have absolutely zero understanding of the task at hand. If you give them problem they have not encountered in training data, they will not solve it. You can however, guide them to resolve the problem, but in that case these get reduced to merely an auto complete. Don't get me wrong - models are getting better and hide very well that they don't understand anything, you can almost get fooled now.
Maybe there aren’t that many new/necessary ideas that can be mined from the fundamental building blocks of software development (languages, syntax, runtimes, static analyses, type checking, etc). Maybe people will continue to innovate by instructing models to build novel things out of those building blocks? Perhaps things we would not have thought of building before due to the effort required without LLM assistance.
It's the opposite. The less competent the average developer the more valuable coding LLMs become (as the only way for those bad developers to generate ok code). Eliminate the good developers and even bad coding LLMs become valuable.
I don't expect them to realise that until some time after it actually happens. When it remains a future hypothetical, it won't be accounted for.
https://en.wikipedia.org/wiki/Profession_(novella)
fwiw 90% of software is reinventing the wheel. 80% of devs have an itch to "rewrite from scratch".
AI will deduplicate all of this
My experience is that 100% of AI devs are reinventing the wheel, most of the time for no better reason than "I can do it" or "not invented here"
I opened LinkedIn today (out of habit) and the first post was someone explaining how much Slack costs and that with AI every company can build their own Slack for $100. So that person decided to build an open source Slack clone using Claude Code. Granted, these were a few sane comments showing good alternatives that have already been built.
But for me it's been a signal that people have no imagination, so they are just burning tokens for no reason.
This is fine. How else do you learn but by taking things apart and rebuilding them? This obsession with productivity is incompatible with onboarding new talent. Having 1000 versions of the same concept is exactly what progress is.
> 1000 versions of the same concept
That sounds beyond wasteful.
It is. Humans are messy.
Why would there be a lack of original ideas? People who are born to code so to speak will do it. Information wants to be free as the saying goes. It only takes one time for an innovation for it to be to copied everywhere.
We don’t need the same volume of developers to have the same or faster speed of innovation.
And conversely if there is stagnation there is a capital opportunity to out compete it and so there will be a human desire to do the work.
Tl;Dr. People like doing stuff and achieving. They will continue to do stuff.
ps it’s too much to claim other people don’t experience creative ideas using AI. You don’t really know that’s true. It hasn’t been my experience as I have had the capability and capacity to complete ideas on my back burner for decades and move onto the next thing.
That’s the big scary point at the crux of all of this - you’ve had decades without the tooling to develop instincts. Nobody knows whether it’s possible to develop instincts with the tooling or what those instincts will look like. Creativity takes a degree of skill to execute on and the concern is that we’re potentially graduating people to painting the ceiling of the Sistine chapel before they’ve even learned to sketch.
At minimum, our current generation of leaders will have to get much better at managing resources and building people up. We have to up our games and build environments where the pursuit of deep understanding is permissible. Unfortunately with the current hiring issues, it’s totally understandable that young developers are scared to take time on tickets.
I can't repair my car, which used to be the hallmark of technical masculine skill in the era of Grease the musical, because mechanical maintenance is not primary in my lifetime. Nor do I have any idea how to manage a farm. I think the kids will be fine. On the other hand, I fear I will not survive once my Internet connection goes out.
Innovation is irrelevant to pushing up this quarter's numbers. No one actually values unique and novel ideas. The only thing that matters is shipping something right now that can make an impact on this quarter's numbers.
Who cares if it's derivative slop or a straight up bootleg of something else so long as the number goes up
It isn’t about training anymore. It is about harnesses.
Just look at new math proofs that will come out, as one example. Exploration vs Exploitation is a thing in AI but you seem to think that human creativity can’t be surpassed by harnesses and prompts like “generate 100 types of possible…”
You’re wrong. What you call creativity is often a manual application of a simple self-prompt that people do.
One can have a loop where AI generates new ideas, rejects some and ranks the rest, then prioritizes. Then spawns workloads and sandboxes to try out and test the most highly ranked ideas. Finally it accretes knowledge into a relational database.
Germans also underestimated USA in WW2, saying their soldiers were superior, and USA just had technology — but USA out produced tanks and machinery and won the war through sheer automation, even if its soldiers were just regular joes and not elite troops.
Back then it was mechanized divisions. Now it is mechanized intelligence.
While Stalin said: Quantity has a quality all its own.
There is no "new ideas" with AI. Claiming the opposite is a fundamental misunderstanding of the technology.
I hear this sentiment a lot but it doesn’t ring true for me.
What is an idea really and what’s your definition of new?
If i get a LLM to spit out, I dunno, a deployment system written in haskell that uses bittorrent or something, none of those bits are new, but certainly there will be unique challenges to solve in the code and it’s a new system.
Where is the line for new? Is it in combining old ideas? If not then does any software have “new” ideas? It’s all combinations of processor instructions after all…
While that’s kind of true in some sense, I think there’s an argument to be made for the contrary: that the mechanism for generating new ideas in humans is not quite as special as we would like to think.
In other words, creativity in humans is arguably just as derivative as in machines.
I think this can be falsified by just considering the history of humanity. It wasn't that long ago that human language literally did not even exist. And our collective knowledge wasn't all that much more than 'poke him with the pointy end'. Somehow we went from that to putting a man on the Moon, unlocking the secrets of the atom, and more. And if you consider how awful we are at retaining/sharing information and just general inefficiencies due to the fact that we're humans and not just logical information processing machines, we did all of this in little more than the blink of an eye. This is something that seems to certainly be rather special.
All that humanity has achieved happened due to the simple loop of identifying a desire/need and finding a way to satisfy it. Also known as reinforcement learning. The only thing that really differentiates humans from machines is... history. We've been learning and passing on our knowledge to successive generations over millennia. Nothing really special there; give the machines a few years to learn and see what happens.
What needs do machines have? What desires do they have?
I’m not claiming an LLM is structurally or functionally equivalent to a human brain. I just said that what we call “creativity” is in fact a very derivative thing.
What I am excited about is the possibility of LLMs to draw conclusions from the last 150years of scientific papers.
There have been lots of instances of knowledge being rediscovered even when it was previously published but sitting on some shelf forgotten. LLMs ability to digest large volumes of data will I think help with this issue.
We will still need to reproduce and verify conclusions but will be interesting to see what might come from this.
i don't think all sides of this discussion agree on what a "new idea" is. i am a very creative person but i've never had a truly original thought and i don't know how having one would be possible
It depends on what layer you look at I think, shoulders of giants and all that..
that's only partially true.
AI can innovate in synthetic-realm of novel ideas, while real-world novelty will remain untouched.
There are different types of novelties
If AI could innovate it wouldn't be a public product. It would be a cash cow. Why give your customers the ability to come up with new and amazing ideas when you can just keep it for yourself and launch a thousand products? USA is a capitalist society. It doesn't share profitable ideas.
And if AI was really about productivity they'd be talking about doing more faster with the same workforce, not reducing the workforce.
if you like, the business model is called Innovation-as-a-Service :)
That's perfectly aligned with capitalistic motivations
What is a "real-world novelty" and what prevents AI from touching it?
Technically all the problems that almost any given business needs to be solved today has already been solved umpteen times over the years. There are no new problems that can't be solved by porting and/or combining old solutions.
"Everything has already been invented." - Some 19th-century scientist who had no imagination to see the wave of technological innovation that was coming.
That's the literal definition of stagnation. That is not compatible with growth.
Also that's not a new idea, that "everything worth inventing/exploring has already been". It's precisely what AI reinforces, and that goes against human nature (and capitalism) as that statement has historically proved.
I think this concern is overblown. AI is an incredible teaching tool. It's probably better for teaching/explaining than for writing code. This will make the next generation of junior devs far more effective than previous generations. Not because they're skipping the fundamentals...because they have a better grasp of the fundamentals due to back-and-forth with infinitely patient AI teachers.
Not in my experience. They just regurgitate code, and juniors don’t know if/why it’s good or bad and consequently can’t field questions on their PR.
“It’s what the LLM said.” - Great. Now go learn it and do it again yourself.
Unless your company is investing in actually teaching your junior devs, this isn't really all that different than the days when jr devs just copied and pasted something out of stack overflow, or blindly copied entire class files around just to change 1 line in what could otherwise have been a shared method. And if your company is actually investing time and resources into teaching your junior devs, then whether they're copying and pasting from stack overflow, from another file in the project or from AI doesn't really matter.
In my experience it is the very rare junior dev that can learn what's good or bad about a given design on their own. Either they needed to be paired with a sr dev to look at things and explain why they might not want to something a given way, or they needed to wind up having to fix the mess they made when their code breaks something. AI doesn't change that.
I always say "own the output". No need to do it by hand but you better damn well research _why_ the AI chose a solution, and what alternatives there are and why not something else, how it works and so on. Ask the AI, ask a seperate agent/model, Google for it, I don't care, but "I don't know the LLM told me" is not acceptable.
For me, the hardest part of software development was learning incantations. These are not interesting, they're conventions that you have to learn to get stuff to work. AI makes this process easier.
If people use AI to generate code they don't understand, that will bite them. But it's an incredibly tool for explaining code and teaching you boring, rote incantations.
This just means you have bad juniors who aren’t interested in learning.
>AI is an incredible teaching tool.
As a junior, my top issue is finding valuable learning material that isn't full of poor or outright wrong information.
In the best and most generous interpretation of your statement, LLM's simply removed my need to search for the information. That doesn't mean it's not of poor quality or outright wrong.
I suspect that the quality is ironically correlated with the expertise of the user (i.e. it is knowledgeable if you are knowledgeable), which puts you in a conundrum (I can report that with a couple decades of experience, LLMs are giving me high quality, correct results, but I can already see that it somehow doesn't work as well for some of my less experienced colleagues. A lot of what I've been doing over the last couple months is trying to find how to make it "just work" for them.).
As a general principle, take advantage of the fact that it can easily generate stuff. If you don't know whether something is true, have it prove it. Make a PoC/test/benchmark to demonstrate what it's saying. Have it pull metrics that you have access to. Add more observability. Create feedback loops (or rather, ask it to create feedback loops). They're very good at reasoning given access to the ground truth, so give them more ability to ground themselves.
They also have fantastic knowledge of public things, but no knowledge of your company, so your instructions should mostly be documentation of what's unique to your company. If it can write an instruction on its own (e.g. how to use git or kubernetes), it is a useless instruction; it already knows that. What it doesn't know is e.g. where your git server is. It also doesn't know what matters to your company: are you a startup trying to find product market fit? Are you an established company that is not allowed to break customer setups? etc. You might even be able to ask it what kinds of questions a senior might ask about how a company/team works when coming into a new job, and then see if you can answer those questions (or find someone who can). In fact, go ask chatgpt:
> What are some questions a senior engineer might ask when coming into a new role to make themselves more effective?
> What are some questions a principle engineer might ask when coming into a new role to make themselves more effective?
> What are some questions an engineering manager might ask when coming into a new role to make themselves more effective?
> What are some questions an engineering director might ask when coming into a new role to make themselves more effective?
Here's a tip from an old timer: read the official docs.
I work a lot with juniors, and they all seem to prefer watching video's. But videos in my opinion are a slow way to gain superficial knowledge.
Do it the hard way and read the official docs, it will be your superpower. Go fast over the easy parts, go slow over the hard parts, it's that simple.
Research [0] from Anthropic about juniors learning to code with AI/without:
>the AI group averaged 50% on the quiz, compared to 67% in the hand-coding group
And why would they do better? There's less incentive to learn because it's so easy to offload thinking to AI.
[0] https://www.anthropic.com/research/AI-assistance-coding-skil...
Objectively speaking, students that use AI score more than a full grade point below their peers not using AI.
AI makes students dumber, not smarter.
This is the dumbest thought that proliferates this website.
Super great that it’s used to pump out tons of code because upper management wants features released even faster than before. I’m sure the junior devs who don’t know a for loop from their ass will be able to learn and understand wtf Claude is shitting out
> AI is an incredible teaching tool. It's probably better for teaching/explaining than for writing code.
It is but how do you teach to people who think their new profession is being a "senior prompt engineer" (with 4 months of experience) and who believe that in 12 months there won't be any programmer left?
A teacher who just gives you the solution isn’t a good teacher.
You can use AI as a teacher but how many will do that?
Highly motivated people will use whatever tools they have to get better at something, whether they have a textbook, the internet or a LLM to use.
The skill of the very top programmers will continue to increase with the advent of new tools.
And how many will not? For mist people it’s just a job to get money, they will put exactly as much effort in it as is necessary to produce an acceptable result
Only for people who wants be taught, this argument keeps coming up again and again but people in general doesn’t want to learn how to fish, they want the fish on a plate ready to eat, so that they can continue scrolling. I see this a lot in juniors, they are solution seekers, not problem solvers, and AI makes this difference a lot worse.
I do agree it’s a great tool, so much better than trying to hope and pray someone on the internet can help you with “I don’t understand this line of code.”
However, it’s got a lot of downsides too.
I see so much creativity coming from young developers I just can’t agree. Yes most developers in the past 20 years who were only chasing big tech money were useless. Good riddance
Ai might bring forward the standardisation we never had. If coding dynamics shift enough then all the opinions about libraries and engines and frameworks might get less focused on readability and more on efficiency and easy composition by Ai.
Security gets outsourced to audited layers and Ai does the stupid boring jobs of gluing them together. Some developers become more specialised and niche, some pivot to product, some pivot to other areas.
There are plenty of people who joined software for the payout and hate it. Plenty of people who grown to hate it over time.
I've been enjoying using it to figure out toy projects but paying an API and depending on a service to code is very sour. I really hope hardware specialises and local models become good enough. Gate keeping development on centralised services would be a loss for everyone and ripe for dystopian outcomes.
It also makes junior devs unobtainable. Because who in their right mind is going to start a career in CS these days?
Idk, anyone who wants to understand and apply the most powerful and transformative technology in human history…?
1. There isn't much to understand about this technology.
2. Unlike in the past, you can't program the technology without having billions in cash.
Copying homework and cheating at exams don't make student learn.
It takes time to become a junior too. Emerging tech landscape could affect skills and knowledge that is expected from entry level job applicants.
I recently read a similar discussion in the context of AI in science and PhD students. And the point the author was making that the goal of having PhD students is NOT to produce academic research, but to train people. I think the same idea applies here. Somebody still needs to train people, and the companies will probably need to ensure that they have resources for that, as there will not be enough senior people for all the tasks.
I understand that but try getting hired in industry as a PhD with that argument.
Back to fuedalism we go!
Is this writeup AI generated?
Every week, I read an article on the consequences of reliable coding agents in SWE industry. All such discussions on HN leads to a fundamental suspicion of the empirical scaling laws of LLMs or the infinite greed and short-sightedness of the market in inflating a bubble. I'm tired.
It's interesting to watch industry after industry hollow itself out from the inside then inevitably die long after all the financial people, investment bankers and management consultants have all cashed their checks.
Steve Jobs famously accurately called this out years ago [1].
Xerox, Boeing, PC manufacturers (who basically created the Taiwanese makers through a series of short-term outsourcing steps), etc. But there are two examples I want to talk about specifically.
First, one lasting impact of the 2008 GFC was that entry-level jobs disappeared. This devastated a generation of millenial college graduates who suddenly had a mountain of student loan debt (thanks to education costs outpacing inflation by a lot) but suddenly no jobs. It became a bit of a joke to poke fun at such people who had a ton of debt and worked as baristas but this was a shallow "analysis". It was really a systemic collapse. Those entry-level workers are your future senior workers and leaders. Those jobs have never come back.
The rise of DVR/TiVo and ultimately streaming brought on a golden age of TV in the 2000s. It was kind of the last hurrah for network shows that produced 22 episodes a year before streamers instead produced 8 episodes every 4 years.
But what made this system work was an ecosystem. Living in LA, Atlanta and a few other places was relatively cheap so aspiring actors and writers and entertainmnet professionals could get by with secon djobs and relatively low income. These became the future headline actors and senior professionals. Background work and odd jobs were sufficient. Background work also taught people how to be on a set.
Studios still had large writing staffs. Some writers would be on set. Those writers were your future producers and showrunners.
Part of what supported all of this was syndication. That is, networks produced shows and basic cable channels would pay to rerun them. Syndicating some shows was incredibly profitable in some cases (eg Seinfeld).
So the streamers came along and stripped things down. They got rid of junior positions. They adopted so-called "mini writing rooms". Those writers didn't tend to ever be on set. The runs were shorter and an 8 episode series couldn't support a writer in the same way a 22 episode series could. The streamers then were largely showing just their own content so residuals and syndication fees just went away.
All of this is short-term thinking. Hollywood has been both a massive industry and a source of American soft power internationally by spreading culture, basically.
I think the software engineering space is going through a similar transformation to what happened to the entertainment industry. A handful of people will do very well. AIs will destroy entry-level jobs and basically destroy that company and industry's future.
I predict in 10-20 years we'll see China totally dominating this space and a bunch of Linkedin "thought leaders" and politicians will be standing around scratchin their heads asking "what happened?"
[1]: https://www.youtube.com/watch?v=K1WrHH-WtaA
I've seen a lot of hot takes on HN but faulting streamers for the inability of Hollywood to adapt to cultural and technological change is gonna require oven mitts. The rise of streaming is a win for the little guy and the viewers against Hollywood's capitalist entertainment oligopoly.
Yes and no. Often times managers are now asking ask Claude code to write it but I want it delivered tomorrow. This leads to us forcing to use LLM generated code without enough time to review or understand it.
I hope you don’t have actual users.
Don't worry, he works for the DoW.
"AI is making junior devs useless" is a dangerous and incorrect conclusion. If this idea is repeated too often, people may start to believe it and even quit studying computer science altogether.
First of all, developers who only learn to code in a short bootcamp are often not well prepared — but that was already true before AI. In the past, many junior developers were students who were learning programming while studying, not just people who took a quick Python course on Udemy.
Instead of declaring junior developers useless, we should raise the standard: learn how to code properly, how to maintain code, understand networks, and build strong foundations in math and computer science. A well-trained junior developer is still extremely valuable and will always be needed.
I assume junior devs can at least search. AI often doesn't even do that. That's why there are things like context7, which in a narrow context helps but not perfect.
There are lots of ambiguous situations where a search and human "inference" can solve that AI still can't.
I can tell the AI to do something, it uses the worst approach, I tell it a better way exists, it says it validated it does not, I link to a GitHub issue saying it can be done via workarounds and it still fails. It's worse for longer tasks where it always shortcuts to if it fails pick a "safe" approach (including not doing it).
Funny enough we need the junior to guide the AI.
Junior devs: you have an oracle you can pester incessantly. Make the most of it so you can learn to detect its mistakes, know when to push back, and what to ask of it. That's when you are in the clear. Juniors who merely parrot the LLM get fired.
It’s gonna take a long time for that to become the norm I think. I really wish I could take 5 years off while everyone figures this out.
You don't have to take 5 years off for it. Just continue same old business (assuming you don't have outside pressure to use this slop-machine) and keep your skills and judgment sharp until the tools and workflow stabilizes, and most of all, the money to fuel this hype runs out.
The reason I say that is because I don’t want to deal with trying to keep things from collapsing around me while people slam in tens of thousands of lines of vibe coded slop.
This is going to be music to deaf ears.
Companies will continue to demand it (I know people working at companies that are literally looking at AI usage as an individual performance metric, puke emoji), and probably 95% of humans using pretty understandable human logic aren’t going to work harder than they need to on purpose.
I wish I had a solution. I think the jury is still out on whether programming will be a dead profession in a short number of years, replaced by technical protect operators.
Problem is not making juniors useless. They kind of are by definition. Problem is that now they have very little chance to become seniors.
"juniors are useless": Maybe y'all should consider updating this hyberbolic language. Nobody is born a "senior developer", so surely all of your training as a "junior" is not useless. There is always a disconnect between what younger people know and what older people expect them to know, so training is required almost universally.
>"juniors are useless. Maybe y'all should consider updating this hyberbolic language"
Don't be so dense. It is a figure of speech. We all were useless at some point. Nothing to be ashamed of
> If I’m reviewing your code and I ask you why you went with a certain approach, and you tell me “the AI suggested it”, I’ve immediately lost confidence in you.
I’ve experienced similar things and so understand the feeling, but this is poor leadership. If someone on your team makes it all the way to a code review and still thinks ‘the AI suggested it’, you failed to train them, failed to set expectations and they have justifiably lost more confidence in you than vice versa.
If we analyze the rest of the article through the lens of weak leadership, it sounds less like an AI problem and more like a corporate leadership problem.
Useless? Where do they expect the senior engineers to come from in the future?
AI made juniors without potential useless, not all juniors.
This is good advice for seniors too.
Eg. When using Ai Deep Research for hard to debug issues, asking for the why makes for a much better response.
You're going to have to do the unthinkable:
Invest in the training of your junior employees.
The cost of generating code is now laughable, so that's not the economic value brought to the table by a junior engineer, or really, any engineer. The value is now generated by knowing what code is good code. You're going to have to have talks, book clubs, hackathons, and the like to get your juniors to know what good code is. Do they know what design patterns are? How about good architecture? If they can't name a few design patterns, you're not investing enough.
Just another silly uninformed take.
This is ridiculous. New developers will learn a completely different skill path from what we learned, and they will get where we are faster than we did.
tl;dr ask why
When I started my career I heard people say almost verbatim "Stack overflow is making junior devs useless", with the idea all we did was copypaste scripts over. The same people failed, and the same people who can use the tools will succeed now.
You definitely did see a difference between people who just copy pasted from stack overflow, and from people with good fundamentals. The uncomfortable truth though, is that the industry didn't need good coders, it needed a bucket load of basic web apps and it needed bums in seats.
I think the irony of AI is going to be that it will make the remaining software jobs properly hard again, and implementers (ex coders) will be able to succeed with even less code knowledge than before.
I'm not sure sure.
I worked under people who started as juniors that way but were politically savvy. Or just ruthless. And pushed their way to the top by stealing projects, lying through their teeth, and other such tactics.
They were slowing down progress because their methods involved sabotaging the progress of others because it might make their own contributions shine a little less.
They were the cause of using libraries like leftPad all through business critical code, and cutting anyone down who dared to simply question why.
These things cause ripples. The smartest and most capable staff leaves, what results is a churn of the same kind.
But hey, they get a trip to Mexico every year and burn through millions every two years. Profit any day now.
I did my first completely vibe coded not looking at a line of code implementation last year and my second this year.
I could care less about why either Claude, Codex or before that a developer was using a for loop or a while loop. I did and do care about architecture.
I’m no more going to review every line of code with AI than I am when I was delegating to more junior developers. I’m going to ask Claude Code about how it implemented something where I know there is an efficient way vs naive way, find and test corner cases via manual and automated tests and do the same for functional and non functional requirements.
The "Junior Trap" is real: if you offload your thinking to Claude or GPT-4, you’re hitting "Done" for the day, but you’re accruing massive Learning Debt. You aren't building the failure-pattern recognition that actually makes an engineer valuable.
In a world where "Code is no longer a skill," the only way to survive is to stop being a "Prompt Operator" and start being a "System Auditor." If you can’t explain the trade-offs of the architectural pattern the AI just gave you, you aren't an engineer, you're just the person holding the screwdriver while the machine builds the house.
Nah, it makes teams useless. Maybe not quite yet, but soon, one engineer will be able to do a few sprint teams' worth of work, and deliver features orders of magnitude faster than a team working in parallel. Yeah, generally at first this will be seniors only. But before long, a junior will be able to come in and learn to manage one sprint team's worth of work under the guidance of a senior and partnered with a PM, and grow the product from there. Long term, I imagine 90% workforce reduction will be the norm. Just about all software is a rinse and repeat of some other software, not much true innovation, so picking and choosing and implementing some other software's feature into your own will start to become trivial single-day projects from start to finish. Hopefully AI creates some new industries that SWEs can roll into, but I'm feeling more doomer every day.
Yeap, and people are still forcing juniors to make small code changes when they should be learning by creating entire apps on their own, deploying them, etc. WITH a senior giving them feedback occasionally. I think people are going to take a while to catch on though, for better or worse....
Yeah, IMO one of the first things we'll see change is more of a migration back to monoliths. Right now adding a feature has to go through multiple teams, a dozen services, a coordinated implementation and deployment schedule, a Byzantine and often manual set of integration tests, etc. Yeah AI can help with that, but the point is that AI doesn't need it. On a monolith, it sucks for dev teams because parallel development at large scale is difficult and other teams' bugs can delay the launch of your unrelated project. Hence, microservices became popular. With AI, development happens so fast that it's largely serial. So there's no real coordination needed. A whole feature is one PR, one set of tests, one app to run locally if you want, one deployment, one thing to look at and roll back if there's a bug. Creates a virtuous cycle all the way up.
I imagine lots of established companies will struggle migrating back to that pattern, but I have to think most new companies will head in that direction, which should let them catch up quickly.
Anyway that's my take. We'll see.