Ultimately, anecdotes and testimonials of a product like this are irrelevant. But the public discourse hasn't caught up with it. People talk about it like it's a new game console or app, giving their positive or negative testimonials, as if this is the correct way to validate the product.
Only rigorous, continual, third party validation that the system is effective and safe would be relevant. It should be evaluated more like a medical treatment.
This gets especially relevant when it gets into an intermediate regime where it can go 10,000 miles without a catastrophic incident. At that level of reliability you can find lots of people who claim "it's driven me around for 2 years without any problem, what are you complaining about?"
10,000 mile per incident fault rate is actually catastrophic. That means the average driver has a serious, life threatening incident every year at an average driving rate. That would be a public safety crisis.
We run into the problem again in the 100,000 mile per incident range. This is still not safe. Yet, that's reliable enough where you can find many people who can potentially get lucky and live their whole life and not see the system cause a catastrophic incident. Yet, it's still 2-5x worse than the average driver.
It can be misleading to directly compare disengagements to actual catastrophic incidents.
The human collision numbers only count actual incidents, and even then only ones which have been reported to insurance/authorities. It doesn't include many minor incidents such as hitting a bollard, or curb rash, or bump-and-run incidents in car parks, and even vehicle-on-vehicle incidents when both parties agree to settle privately. And the number certainly excludes ALL unacceptably close near-misses. There's no good numbers for any of these, but I'd be shocked if minor incidents weren't an of magnitude more common, and near misses another order of magnitude again.
Whereas an FSD disengagement could merely represent the driver's (very reasonable) unwillingness to see if the software will avoid the incident itself. Some disengagements don't represent a safety risk at all, such as when the software is being overly cautious, e.g. at a busy crosswalk. Some disengagements for sure were to avoid a bad situation, though many of these would have been non-catastrophic (such as curbing a wheel) and not a collision which would be included in any human driver collision statistics.
> Only rigorous, continual, third party validation that the system is effective and safe would be relevant. It should be evaluated more like a medical treatment.
100% agreed, and I'll take it one step further - level 3 should be outright banned/illegal.
The reason is it allows blame shifting exactly as what is happening right now. Drivers mentally expected level 4 and legally the company will position the fault, in as much as it can get away with, to be on the driver, effectively level 2.
They're building on a false premise that human equivalent performance using cameras is acceptable.
That's the whole point of AI - when you can think really fast, the world is really slow. You simulate things. Even with lifetimes of data, the cars still will fail in visual scenarios where error bars on ground truth shoot through the roof.
Elon seems to believe his cars will fail in similar ways to humans because they use cameras. False premise. As Waymo scales, human just isn't good enough, except for humans.
So, I agree with what you're saying, but that doesn't matter.
The legal standing doesn't care what tech it is behind it. 1000 monkeys for all it matters. The point is level 3 is the most dangerous level because neither the public nor the manuf properly operates in this space.
If there was actually a rate of one life threatening accident per 10,000 miles with FSD that would be so obvious it would be impossible to hide. So I have to conclude the cars are actually much safer than that.
FSD never drives alone. It's always supervised by another driver legally responsible to correct. More importantly we have no independently verified data about the self driving incidents. Quite the opposite, Tesla repeatedly obscured data or impeded investigations.
I've made this comparison before but student drivers under instructor supervision (with secondary controls) also rarely crash. Are they the best drivers?
I am not a plane pilot but I flew a plane many times while supervised by the pilot. Never took off, never landed, but also never crashed. Am I better than a real pilot or even in any way a competent one?
Above I was talking more generally about full autonomy. I agree the combined human + fsd system can be at least as safe as a human driver, perhaps more, if you have a good driver. As a frequent user of FSD, it's unreliability can be a feature, it constantly reminds me it can't be fully trusted, so I shadow drive and pay full attention. So it's like having a second pair of eyes on the road.
I worry that when it gets to 10,000 mile per incident reliability that it's going to be hard to remind myself I need to pay attention. At which point it becomes a de facto unsupervised system and its reliability falls to that of the autonomous system, rather than the reliability of human + autonomy, an enormous gap.
Of course, I could be wrong. Which is why we need some trusted third party validation of these ideas.
Yeah, I agree with that. There's a potentially dangerous attention gap that could just play into the fundamental weakness of the human brain's ability to pay attention for long periods of time with no interaction. Unfortunately I don't see any possible way to validate this without letting the tech loose. You can't get good data on this without actual driving in real road conditions.
At a certain point you do need to test in real road conditions. However, there is absolutely no need to jump straight from testing in lab conditions and “testing” using unmonitored, untrained end users.
You use professional trained operators with knowledge of the system design and operation using a designed safety plan to minimize prototype risks. At no point should your test plan increase danger to members of the public. Only when you fix problems faster than that test procedure can find do you expand scope.
If you follow the standard automotive pattern, you then expand scope to your untrained, but informed employees using monitored systems. Then untrained employees, informed employees using production systems. Then informed early release customers. Then once you stop being able to find problems regularly at all of those levels do you do a careful monitored release to the general public verifying the safety properties are maintained. Then you finally have a fully released “safe” product.
As time goes on, Tesla's fiasco becomes more and more embarrassing. Waymos are all over the place in the cities they serve, doing pretty much what they're supposed to do. Nuro has some fully autonomous vehicles running around. Baidu's Apollo Go is deployed in 16 cities in China, although they use remote driving as a backup.
Tesla, though, is still hyping a technology that seems to have maxed out years ago.
Not sure if it’s evident in broader statistics yet, but I think that because Tesla got the early adopter market (tech savvy people), they are now losing that same market first.
I had a party at my house a couple months ago, mostly SF tech people. I found the Tesla owners chatting together, and the topic was how much FSD sucks and they don’t trust it.
I asked and no-one said they would buy a Tesla again. Distrust because they felt suckered by FSD was a reason, but several also just said Elon’s behavior was a big negative.
I own six EVs (three cars, one of which is a Tesla, and three motorcycles). My first EV was my Tesla.
We're on the cusp of trading the Tesla in for a Rivian most likely. I should be Tesla's target customer, but instead I'm exactly who you described:
- I don't like the brand. I don't like Elon. I don't like the reputation that the car attaches to me.
- I don't trust the technology. I've gotten two FSD trials, both scared the shit out of me, and I'll never try it again.
- I don't see any compelling developments with Tesla that make me want to buy another. Almost nothing has changed or gotten better in any way that affects me in the last four years.
They should be panicking. The Cybertruck could have been cool, but they managed to turn it into an embarrassment. There are so many alternatives now that are really quite good, and Tesla has spent the last half a decade diddling around with nonsense like the robot and the semi and the Cybertruck and the vaporware roadster instead of making cars for real people that love cars.
Their ~mission was cheap cars for masses. There are plenty of high end EVs out there. It's nuts when smart people compare 7 year old 35k Tesla to a brand new 80k Polestar or 100k Lucid.
> instead of making cars for real people that love cars.
Whoosh. They've been saying Tesla is an AI company for nearly a decade. AI has been propping up entire US economy for last few years. EV bandwagon has left long time ago.
Saying all that I wouldn't mind even cheaper Tesla - small screen, 1 camera instead of 11, fully offline, fully stainless steel, fully open source - basically minimally tech and maximally maintainable and maximum longevity.
> Saying all that I wouldn't mind even cheaper Tesla - small screen, 1 camera instead of 11, fully offline, fully stainless steel, fully open source - basically minimally tech and maximally maintainable and maximum longevity.
What you describe would probably cost more money, not less. The market is small and analog tech is actually more expensive to produce with than digital tech.
They make cars that mostly do their job. They don't make AI that does its job. They're not an AI company, they're a car company pretending they're not a car company.
I'm sure they would be if the stock price had ever showed any signs of being based in reality.
But for now Elon can keep having SpaceX and xAI buy up all the unsold Teslas to make number go up.
If that ever stops working, just spin up a new company with a hyper-inflated valuation and have it acquire Tesla at some made up number. Worked for him once, why not try it again.
And at this point he can get even fraudier, with the worst possible realistic outcome being that he might get forced to pay a relatively small bribe and publicly humiliate himself for Trump a bit.
But there's really no more consequences to any sort of business fraud (for now) as long as you can afford the tribute.
I partially agree. FSD seems fine to me but I wouldn’t buy a second Tesla. Tesla seems to have stopped caring about being a car company that caters to nerds/tech enthusiasts.
Mine has been an extremely well done vehicle and I was (and kind of am) bullish on FSD as a driver assistance technology, but a car is a 6-7 year investment for me and I have big doubts about their direction. They seem to have abandoned the idea of being a car company, instead chasing this robotaxi idea.
Up until 2023/2024 was fine for my 6-7 year car lifecycle. Tesla was really cool when they let you do all sorts of backwards-compatible upgrades, but they seemed to have abandoned that.
I’ve found it incredibly disappointing seeing their flailing direction now.
Rivian seems to still have a lot of the magic that Tesla had. They’re definitely a strong contender for my next vehicle in a year or two.
I just had a thought - a Waymo car costs $200k (maybe more) from a quick google search. YoY returns of $200k on S&P are about 10%, while an Uber driver takes home about $40-$50k - so in terms of cost, they are about 2x-2.5x of each other, with the Waymore likely needing expensive maintenance/support infrastructure, bringing the total much closer.
Which means if Tesla can really build that Cybercab - with an underpowered motor, small battery, plastic body panels, just cameras (which I think they promised to sell under $20k) - they'll be able to hit a business expense level and profitability that Waymo will only be able, in say, 10 years.
Even if you don't want to talk about non-existing hardware, a Model 3's manufacturing cost is surely much lower than a Waymo.
Once (if) they make self driving work at any point in time before Waymo gets to the same level of cost - they'll be the more profitable business.
Not only that, they'll be able to enter markets where the cost of Waymo and what you can charge for taxi rides is so far apart that it doesn't make sense for them - in this sense, they'll have a first mover advantage.
Not to mention, Waymo is moving from Jaguar to Zeekr for its next-gen fleet, meaning 100% import tariffs on those Chinese-built base vehicles before it even begins the expensive retrofit process.
The core problem with Waymo’s model is its lack of (economically rational) scalability. Shipping finished vehicles to a second facility to strip and rebuild them as robotaxis is inherently inefficient, and cannot be made efficient by scaling up. To achieve meaningful volume, Waymo would need to contract an automaker to build finished robotaxis, ideally domestically or where tariffs are sufficiently low.
Obviously Tesla's solution only works if their vision-only strategy bears fruit. Assuming it does (a wildly controversial assumption in this space, but let's go with it for now) the economics are utterly wild. It's difficult to imagine how any competitor could come close to competing on cost or scale. And that's assuming the Model Y, ignoring the as-yet hypothetical Cybercab.
I suppose Alphabet could buy the corpse of Canoo. I suspect that if it had a plausible manufacturing ramp, they would have been snapped up quickly. Automotive-scale manufacturing is a crucible, and it destroys most who attempt it. In fact most die long before they begin frfr.
Waymo cars are basically priceless at this point. As in: the car cost doesn't matter. They've so far spent multiple times their fleet's costs on R&D. The fact that they're getting some pocket cash from paid fares is inconsequential for their bottom line.
Any realistic mass deployment will use cheaper cars, more suitable for taxi service.
The cost isn't an issue at their current scale, perhaps not even one order of magnitude larger, because today's fleet is as much a devkit as it is a consumer product. But that strategy only takes them so far. Cost is a dangerous impediment to scale, and it will be their albatross long before their business model has a chance of becoming profitable, let alone cumulatively profitable.
Tesla's FSD has indeed made significant improvements in the past year (still way behind where it was promised to be even half a decade ago), but they are FAR from being able to operate an actual robotaxi service. Austin is an embarrassment. It seems that tesla believes they can make more money on fooling investors than they can on any core business model
Sorry but while I am a happy Waymo user, this is overblown and just incorrect. Waymo has datacenter oversight. (which, who cares, the product is great).
Tesla FSD 12 -> 13 was a massive jump that happened earlier this year. 14 is still rolling out.
Testing out 13 this weekend, it drove on country roads, noticed a road closure, rerouted, did 7 miles on non divided highways, navigated a town and chose a parking space and parked in it with zero interruptions. It even backed out of the driveway to start the trip. I didn't like the parking job and reparked; other than that, no hands, no feet, fully autonomous. Unlike 12, I had no 'comments' on the driving choices - rotaries were navigated very nicely, road hazards were approached and dealt with safely and properly. It was genuinely good.
Dislike Elon all you want, but Tesla FSD is improving rapidly, and to my experienced eyes adding 9s. Probably two or three more 9s to go, but it's not a maxed out technology.
You lack data to draw this conclusion. The most important factor is deaths per mile, which is sparse, so it requires aggregating data from many drivers before you have enough statistical power.
Qualitative reactions matter. I didn't say it was genuinely safe. I can say that it was smooth, reacted better than most adult drivers to the situations I witnessed, and was significantly better at those things than the prior version. Safety is sparse, but steering, acceleration, decision making happen at like 60hz, and I would say I can make a qualitative judgment on it.
"Trust me, bro, it really works" is these guys' slogan. It's like when you hear of a fad diet, and all the evidence for the diet's effectiveness is anecdotal "It totally worked for me!" testimony.
Where "good" is pretty much identical to having a specific rate of accidents over a specific set of conditions, then in fact data is absolutely necessary to determine that. Lol.
I can imagine it doing fine on highways for a thousand miles. FSD has literally never managed to complete a trip involving city driving for me without disengaging or me having to stop it from doing something illegal. I'm not sure how many attempts I'm supposed to give it. Hell, autopilot doesn't even manage to consistently stay safely in a lane on I-40 between Durham and Asheville.
have you tried fsd 13/14 yet? I just upgraded to 13 from 12 and it's a massssive improvement. Not sure what 14 s like. It's definitely added a 9 in the last year.
I've heard this so many times it's starting to be a meme. The system was claimed to be very capable from the beginning, then every version was a massive improvement, and yet we're always still in very dangerous, and honestly underwhelming territory.
Teslas keep racking up straight line highway miles where every intervention probably counts at most as 1 mile deducted from the total in the stats. Have one cross a busy city without interventions or accidents like a normal human driver is expected to.
It was very capable, and each version has been a big improvement. The first time I rode in a Tesla with FSD back in 2017 I was shocked by how good it was. Self driving tech has advanced so fast that we forget it was considered next to impossible even 15 years ago. You are judging past tech by 2025 standards.
Novelty is enough to look like "shockingly good". You were comparing "no self driving" to "some self driving". A jump from 0 to something always seems big. Standard driver assists were also impressive when they appeared on cars. In the meantime Tesla still makes a lot of claims about safety but doesn't trust the FSD enough to publish transparent, verifiable stats. That speaks louder than any of our anecdotes.
> You are judging past tech by 2025 standards.
That's very presumptuous of you. Every single person I know driving a Tesla told me the FSD (not AP) is bad and they would never trust it to drive without very careful attention. I can tell Teslas on FSD in traffic by the erratic maneuvers that are corrected by the driver, and that's a bad thing.
> Every single person I know driving a Tesla told me the FSD (not AP) is bad
I really don't believe this because everyone I know who drives a Tesla tells me the opposite. I tend to think this is an artifact of people who just irrationally hate Tesla because IRL every negative thing I hear about Teslas comes from people who don't own the cars and hate Elon Musk.
> they would never trust it to drive without very careful attention
Of course, because the product is not designed to drive without human supervision.
> I can tell Teslas on FSD in traffic by the erratic maneuvers that are corrected by the driver, and that's a bad thing.
I don't believe you actually can because I don't notice any difference in the quality of driving between Tesla's or any other car on the road. (In fact the only difference I can notice between drivers of different cars is large trucks). So, again, I write off such statements as more of the same emotionally driven desire to see a pattern were there isn't one.
> this is an artifact of people who just irrationally hate Tesla
> more of the same emotionally driven desire to see a pattern
Don't you find it curious that every opinion you don't like must be from irrational people hating Tesla, but opinions you do like are all rational and objective? It's as if we didn't define the sunk cost fallacy for exactly this. You're a rational person, if Tesla was confident in the numbers wouldn't we have an avalanche of independently verifiable stats? Instead we're here playing this "nuh-uh" games with you pretending you're speaking with an authority nobody else has. Does any other company go to such lengths to bury the evidence of their success? The evidence that supports their claims?
And of course I can tell FSD drivers, literally nobody else on the street will so often brake hard with absolutely no reason, or veer abruptly then correct and straighten out so hard it wobbles, both on highways and in the city. If it's not the car then it must be the drivers but they wouldn't make such irrational moves.
P.S. The internet is full of video evidence of FSD making serious and unjustified mistakes. Every version brings new evidence. How do you explain those? Irrational haters inventing issues? Car misbehaving only for them? Because you see, even if you film 10 times and get the mistake only once it's still very serious.
Yeah, that's the type of feedback I absolutely do believe. Sounds like something someone would say about their car to me IRL. That's basically the standard I apply to internet comments.
For what it's worth, that was not my experience. 12 didn't fix all of the issues I'd heard about its predecessor. 13 didn't fix all the issues I experienced with 12. "Better" isn't enough, it needs to be so good that every issue with the previous generation is resolved. It's never been close to that.
I'm also sick of hearing "have you tried?" And also "it's really improving!"
Maybe the manufacturer should try the next version. And test it. And then try the next version. And test it. And then continue until they have something that actually works.
I tried 12 and 13. My car hasn't gotten 14 and I'm not interested in finding out if it'll get it, because at this point I'm simply not interested even if it's decent.
And there's no guarantee it'll be meaningfully better, if I'm being honest. Why would I believe that all the issues are fixed? For me to be interested, every single problem that I'd encountered needs to be resolved. Why would I even consider accepting anything less? It's not "partial" self driving. An incremental improvement is useless.
Meh. I do a lot of beta testing, and enjoy it. I think it's fun to watch the evolution of this tech over the last seven or eight years. But you don't have to feel that way.
That's not an anecdote, it sounds like an exaggeration bordering flat out lying. A 2000 miles trip "without touching anything" to drive the car is statistically impossible for any reasonable drive (e.g. not endless straight lines on an Australian highway), especially for a Tesla famously known for needing interventions often. Even more advanced autonomous driving systems are far too limited to take arbitrary 2000 miles trip with zero human assistance.
2,000 may be stretching it but it is possible if the driver is trusting enough. Personally many of my disengagements isn't because it is being dangerous, but just sub-optimal such as not driving as aggressive as I want to, not getting into off-ramp lane as early as I like, or just picking weird navigational choices.
Trying to recall but I haven't had a safety involved disengagement in probably a few months across late 13 and 14. I am just one data point and the main criticism I've seen from 14 is: 1) getting rid of fine speed controls in favor of driving style profiles 2) its car and obstacle avoidance being overtuned so it will tap the brakes if, for instance, an upcoming perpendicular car suddenly appears and starts to roll its stop sign.
Personally, I prefer it to be overly protective albeit turn it down slightly and fix issues where it hilariously thinks large clouds of leaves blowing across are obstacles to brake for.
Driver profiles seem like a terrible answer to the question of choosing a maximum speed, both for the driver of the vehicle, and for Tesla — because it shifts the understanding of the car's behaviour from the driver to Tesla. I think it's insane that Tesla would take that risk.
IMHO, it's okay for the driver profiles to affect everything other than max speed, including aggressiveness of acceleration and propensity to change lanes. But since exceeding speed limits is "technically" breaking the law, the default behaviour of FSD should be to strictly obey speed limits, and drivers should be given a set of sliders to manually override speed limits. Perhaps like a graphic EQ with sliders for every 10 MPH where you can manually input decide how many MPH over that limit is acceptable.
This would be an inelegant interface, and intentionally so. Drivers should be fully in control of the decision to exceed the speed limit, and by how much. FSD should drive like a hard-nosed driving instructor unless the driver gives unambiguous permission to do otherwise.
[0] Note that I am describing this based on my understanding of the US environment. I am Australian, and our speed limits are strictly enforced at the posted speed, without exception. On any road, you should expect a fine if going 3—6 km/h [2—4 MPH] and caught by a fixed or mobile camera. This applies literally anywhere, including highways. By contrast in the USA, I understand that 5—10 MPH on highways has been socially normalised, and law enforcement generally disregards it.)
This scam is true for all AI technologies. It only "works" as far as we interpret it as working. LLMs generate text. If it answers our question, we say that the LLM works. If it doesn't, we say that it is "hallucinating".
It's not a scam because it does make you code faster even if you must review everything and possibly correct (either manually or via instruction) some things.
As far as hallucinations go, it is useful as long as its reliability is above a certain (high) percentage.
That's the point, nobody really believes there is an intelligence generating Google results. It is a best-effort based engine. However, people have this belief that ChatGPT has somehow some intelligent engine generating results, which is incorrect. It is only generating statistically good results; if it is true or false depends on what the person using it will do with the results. If it is poetry, for example, it is always true. If it is how to find the cure for cancer it will with very high probability be false. But if you're writing a novel about a scientist finding a cure for cancer, then that same response will be great.
Im sorta beginning to think some LLM/AI stuff is the Wizard of Oz(a fake it before you make it facade).
Like why can an LLM create a nicely designed website for me but asking it to do edits and changes to the design is a complete joke. Lots of the time it creates another brand new design (not what i asked all) and it's attempts at editing it LOL. It makes me think it does no design at all rather it just went and grab one from the ethers of the Internet acting like it created it.
I hear you, but GenAI also gets the opposite fork from people who hate it: It's good result that used GenAI at any point => your prompting and curation and editing is worthless and deserves no credit; it's not good result => that proves AI isn't real intelligence.
As with Marmite, I find it very strange to be surrounded by a very big loud cultural divide where I am firmly in the middle.
Unlike Marmite, I wonder if I'm only in "the middle" because of the extremities on both ends…
> Would at least like to use it for stop and go traffic, which is about the only thing I trust FSD for.
Depends on the type of stop and go driving. Crawling along at 15mph, sure. But the most dangerous driving scenario - whether human or machine is the driver - is a scenario with large variations in speed between vehicles and also limited visibility.
For example suddenly encountering a traffic jam that starts around a blind corner.
That's also the most tiresome part of driving and has the least risk due to low speeds. Easy win for FSD. But for all other cases it becomes a complicated ethical question.
> So, for example, when a Florida driver on Autopilot drops his phone and blows through a stop sign, hitting a car which then hits two pedestrians, killing one, Tesla will claim “this driver was solely at fault.” In that case, a judge agreed that the driver was mostly at fault, but still assigned 33% of blame to Tesla, resulting in a $243 million judgment against the company.
His foot was on the gas though
Looking at this author's other articles, he seems more than a bit unhinged when it comes to Tesla: https://electrek.co/author/jamesondow/ Has Hacker News fallen for clickbait? (Don't answer)
A couple of facts on the Florida case: it was a jury verdict, not a judge. The jury found Tesla 33% at fault for a 2019 Key Largo crash. Damages were $129M compensatory (Tesla responsible for 33% of that) plus $200M punitive, for $243M total.
The driver admitted he looked down after dropping his phone and blew a stop sign; Tesla argues his foot was on the accelerator, but the jury still assigned partial fault because Autopilot was allowed to operate off limited-access highways and the company didn’t do enough to prevent foreseeable misuse. The driver had already settled separately.
is there any blame to be associated to Tesla for its feature? what's the right percent for you? 20%? 10%? 5%? 0%?
If the wheels of the car fell off, whould Tesla have any blame for that? If we had laid wires all along the road to allow for automatic driving, and Tesla's software misread that and caused a crash, would it be to blame?
When is Autopilot safe to use? Is it ever safe to use? Is the fact that people seem to be able to entirely trick the Autopilot to ignore safety attention mechanisms relevant at all?
If we have percentage-based blame then it feels perfectly fine to share the blame here. People buy cars assuming that the features of the car are safe to use to some extent or another.
Maybe it is just 0%. Like cruise control is a thing that exists, right? But I'm not activating cruise control anywhere near any intersection. Tesla calls their thing autopilot, and their other thing FSD, right? Is there nothing there? Maybe there is no blame, but it feels like there's something there.
0%. This is entirely on the driver. He's someone who should spend a few years in prison, and then never be allowed to have a license again.
A foot on the gas overrides braking on autopilot and causes it to flash up a large message up on the screen that "Autopilot will not break / Accelerator pedal is pressed"
They don’t even work well in California. Driving into sun? Car loses all lane visibility and yells at you. Driving late at night? Can’t detect lanes and complains about blinded cameras.
If only there was some kinda technology that didn’t rely on optics that could see in pitch dark or when the sun is shining.
Next up: they put LCD layer with selective dimming over cameras and claim world's first instead of investing in lidar/radar as they should from the start rather than dropping it "coz it is not needed"
According to users, that issue appears to be solved as of FSD v13. The solution may be reliant on the higher quality camera modules shipped with hardware 4.
Technically you kind of get this in Nevada when using Tesla insurance and if you drive 100 % FSD. If you drive manually you are pretty much doxed for random Front collision Warning which is super sensitive
My question about the safety of "FSD" or whatever autonomous cars is -- how much safety reputation can you burn in a new market that offers huge convenience? Which is more important, being first or being best?
Apple's famous for not being first, but being best to a well established market and cleaning up.
People certainly don't remember de havilland, who was first to the market of "passenger jet airliner".
I certainly won't be putting my kids in any of these, and especially not in a vehicle operated by a company with a "devil may care" approach to safety.
Sure, it works fine in Arizona or Texas; let's see it work okay with a snow storm in Boston.
Having driven Tesla FSD and coded with Claude/Codex, it suffers from the exact same issues- Stellar performance in most common contexts, but bizarrely nonsensical behavior sometimes when not.
Which is why I call it "thunking" (clunky thinking) instead of "thinking". And also why it STILL needs constant monitoring by an expert.
No. There's a critical difference between AI as a tool and AI as a product. When you use a tool, the responsibility to verify the output is yours. But when you pay for a product like Tesla's Full Self-Driving, the company assumes liability for its performance. If FSD causes harm, Tesla must be held responsible.
That needs to be adjusted for Tesla putting the Robotaxi "safety driver" back behind the wheel when the self-driving Robotaxi thing didn't really work.
It's like being a loan guarantor. If the loan gets paid off, the lender/borrower get the benefits. You get 0. If the loan goes delinquent, you are on the hook.
I’m still perplexed by the utter lack of regulatory action in both the U.S. and Canada with FSD. Its inconsistency and Tesla’s “seat of the pants” approach to safety have gotten enough people injured to where I’d have expected someone to take notice by now. Don’t get me wrong, when it works, it’s amazing but there’s no way to reliably establish that it works on the same route twice.
This is a fairly weak shower thought, but it was interesting to note (anecdotally) people's propensity to rigidly prioritize safety in some, but not all circumstances.
For example, I've had numerous conversations where people will point to safety rating in vehicles to defend their purchasing decisions. Its simple to understand really, I want the safest car for my family/child etc, that is why I refuse to buy an older, used vehicle or prefer a sedan over SUV. Safety becomes cover for preference and defending trends like expanding pickup truck sizes since the 2000s while there is no safety rating or even objective measure of the efficacy of these self-driving systems.
Hopefully I haven't wasted your time, its just a psychological trend that I think exists.
Shame they don't reduce safety ratings for the bigger vehicles with worse visibility that are more likely to kill pedestrians/bicyclists and cause more respiratory problems than smaller vehicles.
Show me actual stats on the safety vs human drivers and I may actually care. The lack of them makes me think that it isn't actually that dangerous compared to humans.
Agreed. Unfortunately, Tesla’s quite cagey with publishing safety data, far as I can tell. For what it’s worth, both HW4 and HW3 with FSD 13.x and 12.x have put me in dangerous situations enough times to where I no longer trust it.
I mean, obviously. A significant part entire economy runs on plausible deniability. Amazon is a great example of this, by keeping similar products or different vendors in the same bin. Regulations are to keep out competitors, etc.
Ultimately, anecdotes and testimonials of a product like this are irrelevant. But the public discourse hasn't caught up with it. People talk about it like it's a new game console or app, giving their positive or negative testimonials, as if this is the correct way to validate the product.
Only rigorous, continual, third party validation that the system is effective and safe would be relevant. It should be evaluated more like a medical treatment.
This gets especially relevant when it gets into an intermediate regime where it can go 10,000 miles without a catastrophic incident. At that level of reliability you can find lots of people who claim "it's driven me around for 2 years without any problem, what are you complaining about?"
10,000 mile per incident fault rate is actually catastrophic. That means the average driver has a serious, life threatening incident every year at an average driving rate. That would be a public safety crisis.
We run into the problem again in the 100,000 mile per incident range. This is still not safe. Yet, that's reliable enough where you can find many people who can potentially get lucky and live their whole life and not see the system cause a catastrophic incident. Yet, it's still 2-5x worse than the average driver.
It can be misleading to directly compare disengagements to actual catastrophic incidents.
The human collision numbers only count actual incidents, and even then only ones which have been reported to insurance/authorities. It doesn't include many minor incidents such as hitting a bollard, or curb rash, or bump-and-run incidents in car parks, and even vehicle-on-vehicle incidents when both parties agree to settle privately. And the number certainly excludes ALL unacceptably close near-misses. There's no good numbers for any of these, but I'd be shocked if minor incidents weren't an of magnitude more common, and near misses another order of magnitude again.
Whereas an FSD disengagement could merely represent the driver's (very reasonable) unwillingness to see if the software will avoid the incident itself. Some disengagements don't represent a safety risk at all, such as when the software is being overly cautious, e.g. at a busy crosswalk. Some disengagements for sure were to avoid a bad situation, though many of these would have been non-catastrophic (such as curbing a wheel) and not a collision which would be included in any human driver collision statistics.
> Only rigorous, continual, third party validation that the system is effective and safe would be relevant. It should be evaluated more like a medical treatment.
100% agreed, and I'll take it one step further - level 3 should be outright banned/illegal.
The reason is it allows blame shifting exactly as what is happening right now. Drivers mentally expected level 4 and legally the company will position the fault, in as much as it can get away with, to be on the driver, effectively level 2.
They're building on a false premise that human equivalent performance using cameras is acceptable. That's the whole point of AI - when you can think really fast, the world is really slow. You simulate things. Even with lifetimes of data, the cars still will fail in visual scenarios where error bars on ground truth shoot through the roof. Elon seems to believe his cars will fail in similar ways to humans because they use cameras. False premise. As Waymo scales, human just isn't good enough, except for humans.
So, I agree with what you're saying, but that doesn't matter.
The legal standing doesn't care what tech it is behind it. 1000 monkeys for all it matters. The point is level 3 is the most dangerous level because neither the public nor the manuf properly operates in this space.
If there was actually a rate of one life threatening accident per 10,000 miles with FSD that would be so obvious it would be impossible to hide. So I have to conclude the cars are actually much safer than that.
FSD never drives alone. It's always supervised by another driver legally responsible to correct. More importantly we have no independently verified data about the self driving incidents. Quite the opposite, Tesla repeatedly obscured data or impeded investigations.
I've made this comparison before but student drivers under instructor supervision (with secondary controls) also rarely crash. Are they the best drivers?
I am not a plane pilot but I flew a plane many times while supervised by the pilot. Never took off, never landed, but also never crashed. Am I better than a real pilot or even in any way a competent one?
Above I was talking more generally about full autonomy. I agree the combined human + fsd system can be at least as safe as a human driver, perhaps more, if you have a good driver. As a frequent user of FSD, it's unreliability can be a feature, it constantly reminds me it can't be fully trusted, so I shadow drive and pay full attention. So it's like having a second pair of eyes on the road.
I worry that when it gets to 10,000 mile per incident reliability that it's going to be hard to remind myself I need to pay attention. At which point it becomes a de facto unsupervised system and its reliability falls to that of the autonomous system, rather than the reliability of human + autonomy, an enormous gap.
Of course, I could be wrong. Which is why we need some trusted third party validation of these ideas.
Yeah, I agree with that. There's a potentially dangerous attention gap that could just play into the fundamental weakness of the human brain's ability to pay attention for long periods of time with no interaction. Unfortunately I don't see any possible way to validate this without letting the tech loose. You can't get good data on this without actual driving in real road conditions.
At a certain point you do need to test in real road conditions. However, there is absolutely no need to jump straight from testing in lab conditions and “testing” using unmonitored, untrained end users.
You use professional trained operators with knowledge of the system design and operation using a designed safety plan to minimize prototype risks. At no point should your test plan increase danger to members of the public. Only when you fix problems faster than that test procedure can find do you expand scope.
If you follow the standard automotive pattern, you then expand scope to your untrained, but informed employees using monitored systems. Then untrained employees, informed employees using production systems. Then informed early release customers. Then once you stop being able to find problems regularly at all of those levels do you do a careful monitored release to the general public verifying the safety properties are maintained. Then you finally have a fully released “safe” product.
As time goes on, Tesla's fiasco becomes more and more embarrassing. Waymos are all over the place in the cities they serve, doing pretty much what they're supposed to do. Nuro has some fully autonomous vehicles running around. Baidu's Apollo Go is deployed in 16 cities in China, although they use remote driving as a backup.
Tesla, though, is still hyping a technology that seems to have maxed out years ago.
Not sure if it’s evident in broader statistics yet, but I think that because Tesla got the early adopter market (tech savvy people), they are now losing that same market first.
I had a party at my house a couple months ago, mostly SF tech people. I found the Tesla owners chatting together, and the topic was how much FSD sucks and they don’t trust it.
I asked and no-one said they would buy a Tesla again. Distrust because they felt suckered by FSD was a reason, but several also just said Elon’s behavior was a big negative.
I own six EVs (three cars, one of which is a Tesla, and three motorcycles). My first EV was my Tesla.
We're on the cusp of trading the Tesla in for a Rivian most likely. I should be Tesla's target customer, but instead I'm exactly who you described:
- I don't like the brand. I don't like Elon. I don't like the reputation that the car attaches to me.
- I don't trust the technology. I've gotten two FSD trials, both scared the shit out of me, and I'll never try it again.
- I don't see any compelling developments with Tesla that make me want to buy another. Almost nothing has changed or gotten better in any way that affects me in the last four years.
They should be panicking. The Cybertruck could have been cool, but they managed to turn it into an embarrassment. There are so many alternatives now that are really quite good, and Tesla has spent the last half a decade diddling around with nonsense like the robot and the semi and the Cybertruck and the vaporware roadster instead of making cars for real people that love cars.
The execution on the roadster baffles me.
IIRC the deposit was 250K, and I know people who signed up on the first day. Can you imagine a more dedicated fan?
How do you not deliver to that group? How big an own-goal is that?
50k. point definitely remains
Their ~mission was cheap cars for masses. There are plenty of high end EVs out there. It's nuts when smart people compare 7 year old 35k Tesla to a brand new 80k Polestar or 100k Lucid.
> instead of making cars for real people that love cars.
Whoosh. They've been saying Tesla is an AI company for nearly a decade. AI has been propping up entire US economy for last few years. EV bandwagon has left long time ago.
Saying all that I wouldn't mind even cheaper Tesla - small screen, 1 camera instead of 11, fully offline, fully stainless steel, fully open source - basically minimally tech and maximally maintainable and maximum longevity.
> Saying all that I wouldn't mind even cheaper Tesla - small screen, 1 camera instead of 11, fully offline, fully stainless steel, fully open source - basically minimally tech and maximally maintainable and maximum longevity.
What you describe would probably cost more money, not less. The market is small and analog tech is actually more expensive to produce with than digital tech.
I said nothing about analog.
They make cars that mostly do their job. They don't make AI that does its job. They're not an AI company, they're a car company pretending they're not a car company.
> They should be panicking.
I'm sure they would be if the stock price had ever showed any signs of being based in reality.
But for now Elon can keep having SpaceX and xAI buy up all the unsold Teslas to make number go up.
If that ever stops working, just spin up a new company with a hyper-inflated valuation and have it acquire Tesla at some made up number. Worked for him once, why not try it again.
And at this point he can get even fraudier, with the worst possible realistic outcome being that he might get forced to pay a relatively small bribe and publicly humiliate himself for Trump a bit.
But there's really no more consequences to any sort of business fraud (for now) as long as you can afford the tribute.
#WorldLibertyFinancial
The fact that Elon has blown the San Franistan EV market should surprise absolutely no one.
I partially agree. FSD seems fine to me but I wouldn’t buy a second Tesla. Tesla seems to have stopped caring about being a car company that caters to nerds/tech enthusiasts.
Mine has been an extremely well done vehicle and I was (and kind of am) bullish on FSD as a driver assistance technology, but a car is a 6-7 year investment for me and I have big doubts about their direction. They seem to have abandoned the idea of being a car company, instead chasing this robotaxi idea.
Up until 2023/2024 was fine for my 6-7 year car lifecycle. Tesla was really cool when they let you do all sorts of backwards-compatible upgrades, but they seemed to have abandoned that.
I’ve found it incredibly disappointing seeing their flailing direction now.
Rivian seems to still have a lot of the magic that Tesla had. They’re definitely a strong contender for my next vehicle in a year or two.
Rivian is definitely up and coming. The increase of them around my neighborhood has been very noticeable over the past 12 months.
I just had a thought - a Waymo car costs $200k (maybe more) from a quick google search. YoY returns of $200k on S&P are about 10%, while an Uber driver takes home about $40-$50k - so in terms of cost, they are about 2x-2.5x of each other, with the Waymore likely needing expensive maintenance/support infrastructure, bringing the total much closer.
Which means if Tesla can really build that Cybercab - with an underpowered motor, small battery, plastic body panels, just cameras (which I think they promised to sell under $20k) - they'll be able to hit a business expense level and profitability that Waymo will only be able, in say, 10 years.
Even if you don't want to talk about non-existing hardware, a Model 3's manufacturing cost is surely much lower than a Waymo.
Once (if) they make self driving work at any point in time before Waymo gets to the same level of cost - they'll be the more profitable business.
Not only that, they'll be able to enter markets where the cost of Waymo and what you can charge for taxi rides is so far apart that it doesn't make sense for them - in this sense, they'll have a first mover advantage.
Not to mention, Waymo is moving from Jaguar to Zeekr for its next-gen fleet, meaning 100% import tariffs on those Chinese-built base vehicles before it even begins the expensive retrofit process.
The core problem with Waymo’s model is its lack of (economically rational) scalability. Shipping finished vehicles to a second facility to strip and rebuild them as robotaxis is inherently inefficient, and cannot be made efficient by scaling up. To achieve meaningful volume, Waymo would need to contract an automaker to build finished robotaxis, ideally domestically or where tariffs are sufficiently low.
Obviously Tesla's solution only works if their vision-only strategy bears fruit. Assuming it does (a wildly controversial assumption in this space, but let's go with it for now) the economics are utterly wild. It's difficult to imagine how any competitor could come close to competing on cost or scale. And that's assuming the Model Y, ignoring the as-yet hypothetical Cybercab.
I suppose Alphabet could buy the corpse of Canoo. I suspect that if it had a plausible manufacturing ramp, they would have been snapped up quickly. Automotive-scale manufacturing is a crucible, and it destroys most who attempt it. In fact most die long before they begin frfr.
Waymo cars are basically priceless at this point. As in: the car cost doesn't matter. They've so far spent multiple times their fleet's costs on R&D. The fact that they're getting some pocket cash from paid fares is inconsequential for their bottom line.
Any realistic mass deployment will use cheaper cars, more suitable for taxi service.
The cost isn't an issue at their current scale, perhaps not even one order of magnitude larger, because today's fleet is as much a devkit as it is a consumer product. But that strategy only takes them so far. Cost is a dangerous impediment to scale, and it will be their albatross long before their business model has a chance of becoming profitable, let alone cumulatively profitable.
Tesla's FSD has indeed made significant improvements in the past year (still way behind where it was promised to be even half a decade ago), but they are FAR from being able to operate an actual robotaxi service. Austin is an embarrassment. It seems that tesla believes they can make more money on fooling investors than they can on any core business model
Sorry but while I am a happy Waymo user, this is overblown and just incorrect. Waymo has datacenter oversight. (which, who cares, the product is great).
Tesla FSD 12 -> 13 was a massive jump that happened earlier this year. 14 is still rolling out.
Testing out 13 this weekend, it drove on country roads, noticed a road closure, rerouted, did 7 miles on non divided highways, navigated a town and chose a parking space and parked in it with zero interruptions. It even backed out of the driveway to start the trip. I didn't like the parking job and reparked; other than that, no hands, no feet, fully autonomous. Unlike 12, I had no 'comments' on the driving choices - rotaries were navigated very nicely, road hazards were approached and dealt with safely and properly. It was genuinely good.
Dislike Elon all you want, but Tesla FSD is improving rapidly, and to my experienced eyes adding 9s. Probably two or three more 9s to go, but it's not a maxed out technology.
Qualitative reactions matter. I didn't say it was genuinely safe. I can say that it was smooth, reacted better than most adult drivers to the situations I witnessed, and was significantly better at those things than the prior version. Safety is sparse, but steering, acceleration, decision making happen at like 60hz, and I would say I can make a qualitative judgment on it.
Completely worthless comment. You don't need data to say something is a good product.
Data is needed to know if it's safe. For self-driving software, safety is the #1 criteria for whether it's a good product.
"Trust me, bro, it really works" is these guys' slogan. It's like when you hear of a fad diet, and all the evidence for the diet's effectiveness is anecdotal "It totally worked for me!" testimony.
Where "good" is pretty much identical to having a specific rate of accidents over a specific set of conditions, then in fact data is absolutely necessary to determine that. Lol.
And yet it drives me on 2000 mile car trips without touching anything. If you ignore the hype and look at the actual product, it’s fine.
I can imagine it doing fine on highways for a thousand miles. FSD has literally never managed to complete a trip involving city driving for me without disengaging or me having to stop it from doing something illegal. I'm not sure how many attempts I'm supposed to give it. Hell, autopilot doesn't even manage to consistently stay safely in a lane on I-40 between Durham and Asheville.
have you tried fsd 13/14 yet? I just upgraded to 13 from 12 and it's a massssive improvement. Not sure what 14 s like. It's definitely added a 9 in the last year.
> have you tried
I've heard this so many times it's starting to be a meme. The system was claimed to be very capable from the beginning, then every version was a massive improvement, and yet we're always still in very dangerous, and honestly underwhelming territory.
Teslas keep racking up straight line highway miles where every intervention probably counts at most as 1 mile deducted from the total in the stats. Have one cross a busy city without interventions or accidents like a normal human driver is expected to.
It was very capable, and each version has been a big improvement. The first time I rode in a Tesla with FSD back in 2017 I was shocked by how good it was. Self driving tech has advanced so fast that we forget it was considered next to impossible even 15 years ago. You are judging past tech by 2025 standards.
Novelty is enough to look like "shockingly good". You were comparing "no self driving" to "some self driving". A jump from 0 to something always seems big. Standard driver assists were also impressive when they appeared on cars. In the meantime Tesla still makes a lot of claims about safety but doesn't trust the FSD enough to publish transparent, verifiable stats. That speaks louder than any of our anecdotes.
> You are judging past tech by 2025 standards.
That's very presumptuous of you. Every single person I know driving a Tesla told me the FSD (not AP) is bad and they would never trust it to drive without very careful attention. I can tell Teslas on FSD in traffic by the erratic maneuvers that are corrected by the driver, and that's a bad thing.
> Every single person I know driving a Tesla told me the FSD (not AP) is bad
I really don't believe this because everyone I know who drives a Tesla tells me the opposite. I tend to think this is an artifact of people who just irrationally hate Tesla because IRL every negative thing I hear about Teslas comes from people who don't own the cars and hate Elon Musk.
> they would never trust it to drive without very careful attention
Of course, because the product is not designed to drive without human supervision.
> I can tell Teslas on FSD in traffic by the erratic maneuvers that are corrected by the driver, and that's a bad thing.
I don't believe you actually can because I don't notice any difference in the quality of driving between Tesla's or any other car on the road. (In fact the only difference I can notice between drivers of different cars is large trucks). So, again, I write off such statements as more of the same emotionally driven desire to see a pattern were there isn't one.
> I really don't believe this
> I don't believe
> this is an artifact of people who just irrationally hate Tesla
> more of the same emotionally driven desire to see a pattern
Don't you find it curious that every opinion you don't like must be from irrational people hating Tesla, but opinions you do like are all rational and objective? It's as if we didn't define the sunk cost fallacy for exactly this. You're a rational person, if Tesla was confident in the numbers wouldn't we have an avalanche of independently verifiable stats? Instead we're here playing this "nuh-uh" games with you pretending you're speaking with an authority nobody else has. Does any other company go to such lengths to bury the evidence of their success? The evidence that supports their claims?
And of course I can tell FSD drivers, literally nobody else on the street will so often brake hard with absolutely no reason, or veer abruptly then correct and straighten out so hard it wobbles, both on highways and in the city. If it's not the car then it must be the drivers but they wouldn't make such irrational moves.
P.S. The internet is full of video evidence of FSD making serious and unjustified mistakes. Every version brings new evidence. How do you explain those? Irrational haters inventing issues? Car misbehaving only for them? Because you see, even if you film 10 times and get the mistake only once it's still very serious.
> I really don't believe this because everyone I know who drives a Tesla tells me the opposite.
I mean, I love my Tesla autopilot. It made my cross-country trips so much more enjoyable. I have several thousand hours on autopilot at this point.
That being said, I don't use it on regular city streets. Because it's just bad, in all kinds of ways. "Full self-driving" it is not.
Yeah, that's the type of feedback I absolutely do believe. Sounds like something someone would say about their car to me IRL. That's basically the standard I apply to internet comments.
For what it's worth, that was not my experience. 12 didn't fix all of the issues I'd heard about its predecessor. 13 didn't fix all the issues I experienced with 12. "Better" isn't enough, it needs to be so good that every issue with the previous generation is resolved. It's never been close to that.
I'm also sick of hearing "have you tried?" And also "it's really improving!"
Maybe the manufacturer should try the next version. And test it. And then try the next version. And test it. And then continue until they have something that actually works.
I tried 12 and 13. My car hasn't gotten 14 and I'm not interested in finding out if it'll get it, because at this point I'm simply not interested even if it's decent.
And there's no guarantee it'll be meaningfully better, if I'm being honest. Why would I believe that all the issues are fixed? For me to be interested, every single problem that I'd encountered needs to be resolved. Why would I even consider accepting anything less? It's not "partial" self driving. An incremental improvement is useless.
Meh. I do a lot of beta testing, and enjoy it. I think it's fun to watch the evolution of this tech over the last seven or eight years. But you don't have to feel that way.
Ancedotes are nice. Miles per disengagement stats would be better.
That's not an anecdote, it sounds like an exaggeration bordering flat out lying. A 2000 miles trip "without touching anything" to drive the car is statistically impossible for any reasonable drive (e.g. not endless straight lines on an Australian highway), especially for a Tesla famously known for needing interventions often. Even more advanced autonomous driving systems are far too limited to take arbitrary 2000 miles trip with zero human assistance.
13 is quite good. 14 is even better.
2,000 may be stretching it but it is possible if the driver is trusting enough. Personally many of my disengagements isn't because it is being dangerous, but just sub-optimal such as not driving as aggressive as I want to, not getting into off-ramp lane as early as I like, or just picking weird navigational choices.
Trying to recall but I haven't had a safety involved disengagement in probably a few months across late 13 and 14. I am just one data point and the main criticism I've seen from 14 is: 1) getting rid of fine speed controls in favor of driving style profiles 2) its car and obstacle avoidance being overtuned so it will tap the brakes if, for instance, an upcoming perpendicular car suddenly appears and starts to roll its stop sign.
Personally, I prefer it to be overly protective albeit turn it down slightly and fix issues where it hilariously thinks large clouds of leaves blowing across are obstacles to brake for.
Driver profiles seem like a terrible answer to the question of choosing a maximum speed, both for the driver of the vehicle, and for Tesla — because it shifts the understanding of the car's behaviour from the driver to Tesla. I think it's insane that Tesla would take that risk.
IMHO, it's okay for the driver profiles to affect everything other than max speed, including aggressiveness of acceleration and propensity to change lanes. But since exceeding speed limits is "technically" breaking the law, the default behaviour of FSD should be to strictly obey speed limits, and drivers should be given a set of sliders to manually override speed limits. Perhaps like a graphic EQ with sliders for every 10 MPH where you can manually input decide how many MPH over that limit is acceptable.
This would be an inelegant interface, and intentionally so. Drivers should be fully in control of the decision to exceed the speed limit, and by how much. FSD should drive like a hard-nosed driving instructor unless the driver gives unambiguous permission to do otherwise.
[0] Note that I am describing this based on my understanding of the US environment. I am Australian, and our speed limits are strictly enforced at the posted speed, without exception. On any road, you should expect a fine if going 3—6 km/h [2—4 MPH] and caught by a fixed or mobile camera. This applies literally anywhere, including highways. By contrast in the USA, I understand that 5—10 MPH on highways has been socially normalised, and law enforcement generally disregards it.)
> 2,000 may be stretching it but it is possible if the driver is trusting enough.
Yeah, trust and a lot of creative accounting of what constitutes successfully driving by itself for that long.
Would you put your child in one and let it cross a large city 100 times unattended?
That also doesn't matter much as highways pump those stats up.
Kind of similar to Agentic coding. The code works? Yay, well done, AI. It't doesn't? You didn't prompt it correctly.
This scam is true for all AI technologies. It only "works" as far as we interpret it as working. LLMs generate text. If it answers our question, we say that the LLM works. If it doesn't, we say that it is "hallucinating".
It's not a scam because it does make you code faster even if you must review everything and possibly correct (either manually or via instruction) some things.
As far as hallucinations go, it is useful as long as its reliability is above a certain (high) percentage.
I actually tried to come up with a "perceived utility" function as a function of reliability: U(r)=Umax ⋅e^(−k(100−r)^n) with k=0.025 and n=1.5 is the best I came up with, plotted here: https://imgur.com/gallery/reliability-utility-function-u-r-u...
How is that any different from Googling something and believing any of the highly-SEO optimized results that pollute the front page?
That's the point, nobody really believes there is an intelligence generating Google results. It is a best-effort based engine. However, people have this belief that ChatGPT has somehow some intelligent engine generating results, which is incorrect. It is only generating statistically good results; if it is true or false depends on what the person using it will do with the results. If it is poetry, for example, it is always true. If it is how to find the cure for cancer it will with very high probability be false. But if you're writing a novel about a scientist finding a cure for cancer, then that same response will be great.
So "statistics" are enough to take gold at IMO?
Search engines didn't need $500B and growing in CAPEX.
Im sorta beginning to think some LLM/AI stuff is the Wizard of Oz(a fake it before you make it facade).
Like why can an LLM create a nicely designed website for me but asking it to do edits and changes to the design is a complete joke. Lots of the time it creates another brand new design (not what i asked all) and it's attempts at editing it LOL. It makes me think it does no design at all rather it just went and grab one from the ethers of the Internet acting like it created it.
> it does no design at all rather it just went and grab one from the ethers of the Internet
Bingo. It just "remembers" one of the many designs it has seen before.
"Trust me, bro, it works" has become kind of a theme song to these guys.
Saved 20 minutes of your time? Slack goes wild.
Wasted 1 hour each of your 5 co-workers who ended up reviewing unusable slop? Silence.
I hear you, but GenAI also gets the opposite fork from people who hate it: It's good result that used GenAI at any point => your prompting and curation and editing is worthless and deserves no credit; it's not good result => that proves AI isn't real intelligence.
As with Marmite, I find it very strange to be surrounded by a very big loud cultural divide where I am firmly in the middle.
Unlike Marmite, I wonder if I'm only in "the middle" because of the extremities on both ends…
Apple did it successfully, it's a marvelous ecosystem, but when there's an issue, it's because you're not holding it the correct way.
Agentic code doesn’t kill people. Tesla drivers who think they own a Waymo do.
We're just waiting for AI code in a Therac-25 type device.
> However, at this week’s shareholder meeting, Musk stated that Tesla may allow “texting and driving” within “a month or two.”
The family of the first person killed by that will know who to sue for a punitive trillion dollars.
Would at least like to use it for stop and go traffic, which is about the only thing I trust FSD for.
> Would at least like to use it for stop and go traffic, which is about the only thing I trust FSD for.
Depends on the type of stop and go driving. Crawling along at 15mph, sure. But the most dangerous driving scenario - whether human or machine is the driver - is a scenario with large variations in speed between vehicles and also limited visibility.
For example suddenly encountering a traffic jam that starts around a blind corner.
Dangerous for some humans. If a self-driving system is taking a blind corner, I would expect it to be approaching it at a more appropriate speed.
Just pull over if you have things to text that can’t wait
Actually just wait. Pulling over on a busy road creates its own hazards.
That's also the most tiresome part of driving and has the least risk due to low speeds. Easy win for FSD. But for all other cases it becomes a complicated ethical question.
Sorry, but disabling autopilot/FSD to use your phone is much crazier.
These are, after all, the only options available to a Tesla driver!
> So, for example, when a Florida driver on Autopilot drops his phone and blows through a stop sign, hitting a car which then hits two pedestrians, killing one, Tesla will claim “this driver was solely at fault.” In that case, a judge agreed that the driver was mostly at fault, but still assigned 33% of blame to Tesla, resulting in a $243 million judgment against the company.
His foot was on the gas though
Looking at this author's other articles, he seems more than a bit unhinged when it comes to Tesla: https://electrek.co/author/jamesondow/ Has Hacker News fallen for clickbait? (Don't answer)
A couple of facts on the Florida case: it was a jury verdict, not a judge. The jury found Tesla 33% at fault for a 2019 Key Largo crash. Damages were $129M compensatory (Tesla responsible for 33% of that) plus $200M punitive, for $243M total.
The driver admitted he looked down after dropping his phone and blew a stop sign; Tesla argues his foot was on the accelerator, but the jury still assigned partial fault because Autopilot was allowed to operate off limited-access highways and the company didn’t do enough to prevent foreseeable misuse. The driver had already settled separately.
Also check the author's Bluesky: https://bsky.app/profile/jamesondow.bsky.social
It can't be healthy to be so obsessed with something/someone you dislike.
is there any blame to be associated to Tesla for its feature? what's the right percent for you? 20%? 10%? 5%? 0%?
If the wheels of the car fell off, whould Tesla have any blame for that? If we had laid wires all along the road to allow for automatic driving, and Tesla's software misread that and caused a crash, would it be to blame?
When is Autopilot safe to use? Is it ever safe to use? Is the fact that people seem to be able to entirely trick the Autopilot to ignore safety attention mechanisms relevant at all?
If we have percentage-based blame then it feels perfectly fine to share the blame here. People buy cars assuming that the features of the car are safe to use to some extent or another.
Maybe it is just 0%. Like cruise control is a thing that exists, right? But I'm not activating cruise control anywhere near any intersection. Tesla calls their thing autopilot, and their other thing FSD, right? Is there nothing there? Maybe there is no blame, but it feels like there's something there.
0%. This is entirely on the driver. He's someone who should spend a few years in prison, and then never be allowed to have a license again.
A foot on the gas overrides braking on autopilot and causes it to flash up a large message up on the screen that "Autopilot will not break / Accelerator pedal is pressed"
One reason I wouldn't trust Tesla's FSD: Sun glare: https://www.reddit.com/r/TeslaFSD/comments/1huzgtu/tesla_fsd...
Drivers have tackled this problem by wearing polaroid sunglasses.
I really hope someone asks Tesla how they plan to solve the Sun glare issue.
I'm sure they'll figure this out right around the time they test the cars in places that have more than one season.
They don’t even work well in California. Driving into sun? Car loses all lane visibility and yells at you. Driving late at night? Can’t detect lanes and complains about blinded cameras.
If only there was some kinda technology that didn’t rely on optics that could see in pitch dark or when the sun is shining.
Next up: they put LCD layer with selective dimming over cameras and claim world's first instead of investing in lidar/radar as they should from the start rather than dropping it "coz it is not needed"
According to users, that issue appears to be solved as of FSD v13. The solution may be reliant on the higher quality camera modules shipped with hardware 4.
Will Tesla recall the old defective cars?
For people who paid for FSD, I think they should.
So why did it ever ship like that?
Why was that OK? Why was it safe to let people use like that without informing them?
When people are so taken by your brand that they volunteer their money and lives to test your half-baked product, why would you stop them?
Because all cars ship with design flaws. Basically any model of car out there is known for its particular parts that break often.
I have hw4. Not fixed.
Tesla's FSD will be good enough when you get an insurance discount for using it.
Technically you kind of get this in Nevada when using Tesla insurance and if you drive 100 % FSD. If you drive manually you are pretty much doxed for random Front collision Warning which is super sensitive
My question about the safety of "FSD" or whatever autonomous cars is -- how much safety reputation can you burn in a new market that offers huge convenience? Which is more important, being first or being best?
Apple's famous for not being first, but being best to a well established market and cleaning up.
People certainly don't remember de havilland, who was first to the market of "passenger jet airliner".
I certainly won't be putting my kids in any of these, and especially not in a vehicle operated by a company with a "devil may care" approach to safety.
Sure, it works fine in Arizona or Texas; let's see it work okay with a snow storm in Boston.
Sounds like a problem with all of AI.
Having driven Tesla FSD and coded with Claude/Codex, it suffers from the exact same issues- Stellar performance in most common contexts, but bizarrely nonsensical behavior sometimes when not.
Which is why I call it "thunking" (clunky thinking) instead of "thinking". And also why it STILL needs constant monitoring by an expert.
isn't that the entire "AI" experience, in a nutshell?
No. There's a critical difference between AI as a tool and AI as a product. When you use a tool, the responsibility to verify the output is yours. But when you pay for a product like Tesla's Full Self-Driving, the company assumes liability for its performance. If FSD causes harm, Tesla must be held responsible.
The modern, "you're holding it wrong" approach to product flaws.
And by the way - I have heard big tech folks repeat that phrase, not really understanding the moral of that Steve Jobs story.
List of predictions for autonomous Tesla vehicles by Elon Musk: https://en.wikipedia.org/wiki/List_of_predictions_for_autono...
That needs to be adjusted for Tesla putting the Robotaxi "safety driver" back behind the wheel when the self-driving Robotaxi thing didn't really work.
I can't believe they are giving credit for "robotaxi" launching with human drivers. Yellow Cab had those in the '50s.
Funnily enough, not an article on grokipedia
I love TACC. I occasionally use AutoSteer. But I tried FSD and it was much more stressful than driving.
It's like being a loan guarantor. If the loan gets paid off, the lender/borrower get the benefits. You get 0. If the loan goes delinquent, you are on the hook.
I’m still perplexed by the utter lack of regulatory action in both the U.S. and Canada with FSD. Its inconsistency and Tesla’s “seat of the pants” approach to safety have gotten enough people injured to where I’d have expected someone to take notice by now. Don’t get me wrong, when it works, it’s amazing but there’s no way to reliably establish that it works on the same route twice.
This is a fairly weak shower thought, but it was interesting to note (anecdotally) people's propensity to rigidly prioritize safety in some, but not all circumstances.
For example, I've had numerous conversations where people will point to safety rating in vehicles to defend their purchasing decisions. Its simple to understand really, I want the safest car for my family/child etc, that is why I refuse to buy an older, used vehicle or prefer a sedan over SUV. Safety becomes cover for preference and defending trends like expanding pickup truck sizes since the 2000s while there is no safety rating or even objective measure of the efficacy of these self-driving systems.
Hopefully I haven't wasted your time, its just a psychological trend that I think exists.
Shame they don't reduce safety ratings for the bigger vehicles with worse visibility that are more likely to kill pedestrians/bicyclists and cause more respiratory problems than smaller vehicles.
It's weird, it's essentially allowing unlicensed driver on roads
Show me actual stats on the safety vs human drivers and I may actually care. The lack of them makes me think that it isn't actually that dangerous compared to humans.
Agreed. Unfortunately, Tesla’s quite cagey with publishing safety data, far as I can tell. For what it’s worth, both HW4 and HW3 with FSD 13.x and 12.x have put me in dangerous situations enough times to where I no longer trust it.
Sounds a lot like chickenization.
thankless works and againts all odds
wonder what driving force behind this, because at somepoint money didnt matters
I mean, obviously. A significant part entire economy runs on plausible deniability. Amazon is a great example of this, by keeping similar products or different vendors in the same bin. Regulations are to keep out competitors, etc.
When stock goes up musk gets money, when it goes down people who short will be the reason !