I can believe this. A lot of my google search usage now is something like:
> "what is the type of wrench called for getting up into tight spaces"
> AI search gives me an overview of wrench types (I was looking for "basin wrench")
> new search "basin wrench amazon"
> new search "basin wrench lowes"
> maps.google.com "lowes"
Notably, the information I was looking for was general knowledge. The only people "losing out" here are people running SEO-spammish websites that themselves (at this point) are basically hosting LLM-generated answers for me to find. These websites don't really need to exist now. I'm happy to funnel 100% of my traffic to websites that are representing real companies offering real services/info (ship me a wrench, sell me a wrench, show me a video on how to use the wrench, etc).
> The only people "losing out" here are people running SEO-spammish websites that themselves (at this point) are basically hosting LLM-generated answers for me to find.
And anybody who creates original content and wishes -- not just to be paid for that content -- but for people to actually see that content and engage with it. IOW the very people who fed the LLM revolution.
> The only people "losing out" here are people running SEO-spammish websites that themselves (at this point) are basically hosting webpages containing LLM-generated answers for me to find.
Agreed. The web will be better off for everyone if these sites die out. Google is what brought these into existence in the first place, so I find it funny Google is now going to be one of the ones helping to kill them. Almost like they accidentally realized SEO got out of control so they have to fix their mistake.
At one point these SEO pages were in fact providing a real service, and you could view them as a sort of "manual", prototypical, distributed form of AI. Millions of people trying to understand what information was valuable and host webpages to satisfy the demand for that info, and get rewarded for doing so. It obviously went too far, but at one point, it did make sense to allow these websites to proliferate. I know without AI, I probably just would have clicked on the first link that said "types of wrenches" and read a little bit. I probably would have gotten my answer, it just wouldn't have been quite as laser-targeted to my exact question.
True, the early days these sites were genuinely helpful. The monetization model was a little different though which is what I think kept them useful. You'd use the content just to drive traffic, which would result in ad clicks on your banner ads, etc.
Then "content marketing" took over, and the content itself was now also used to sell a product or service, sort of an early form of influencer marketing and that is when I think it all started to go down hill. We stopped seeing the more in depth content which actually taught something, and more surface level keywords that were just used to drive you to their product/service.
OTOH, the early web was also full of niche forums, most viewable without an account and indexable, of about any topic you could imagine where you could interact with knowledgeable folks in that niche. Google would have been more helpful to users by surfacing more of those forums vs. the blogs.
Those forums are IMO the real loss here. Communities have moved into discord, or another closed platform that doesn't appear on the web, and many that require accounts or even invitations to just view read only.
Hard disagree. I put a great deal of work into my website, putting hard-earned information on the internet for the first time. Now Google reaps all the value I create without as much as a "thank you".
Unfortunately the new victims in the system of LLMs and the way they distribute knowledge is the specialized content website.
Now an LLM just knows all the content you painstakingly gathered on your site. (It could also be, and is likely that it was also collected from other hard to find sites across the internet).
The original web killed off the value of a certain kind of knowledge (encyclopedias, etc.) and LLMs will do the same.
There are plenty of places to place the blame, but this is a function of any LLM, and a funcamental way LLMs work, not just a problem created and profited from by Google- for example the open-weight models, where no-one is actually profiting directly.
> There are plenty of places to place the blame, but this is a function of any LLM, and a funcamental way LLMs work, not just a problem created and profited from by Google- for example the open-weight models, where no-one is actually profiting directly.
First time learning that scraping and training on data that they have often been explicitly disallowed to obtain for free or for that purpose by the rights holders is "fundamental to how LLMs work". If not, then there is no reason those who gathered the information wouldn't stand to profit by selling LLMs this data.
We killed forums and now were killing specialty websites. Soon we will get right back to pre internet levels of knowledge. Ai will be able to find all the stuff that was easily available in encyclopedias and text books and none of the stuff that made 90s-2000 internet great.
The Internet had plenty of specialty knowledge before "Gathering that knowledge, SEOing it, and stuffing ads all over it" became a viable business model, and it will have plenty of specialty knowledge after this era is over.
Specialty knowledge can also be a guy reviewing air purifiers with obsessive care. It can be someone writing detailed travel guides or in my case an essential resource for immigrants. There were a lot of good websites, and if you could not find them, blame Google for letting its search engine go to the dogs.
Gathering that knowledge is work, and if anyone should capture that value, it's the people doing the work. Seeing bug tech slurp it all up, insert inself in the middle and capture all the value is heartbreaking.
I hate that AI is destroying the economics of the independent web, and people cheer for it because they landed on Forbes one too many times. It's insulting to all the people who did their job right, and still get their work slurped up by an uncaring corporation with no way to stop them.
> There are plenty of places to place the blame, but this is a function of any LLM, and a funcamental way LLMs work, not just a problem created and profited from by Google- for example the open-weight models, where no-one is actually profiting directly.
One problem with monopolies is their massive multiplicative effects on otherwise manageable problems.
Maybe (I can't emphasize enough that I say this with a high degree of skepticism) they'll come up with a way of doing cited/credited/paid work where the LLM will include who "taught" it something.
Eg: instead of writing a blog post, you'll submit a knowledge article to an AI provider and that'll go into the AI's training set and it'll know "you" told it. And maybe (even more skeptical) pay you for it.
Again: highest degree of skepticism, but at the same time, that's the only way I could see people continuing to write content that teaches anything.
If I gave my surplus eggs for free, I'd be annoyed if they ended up at the Best Western's buffet. These eggs were meant for my neighbours.
There is a difference between working for free for your community and working for free for a trillion dollar company's investors. Doubly so when you strip consent or attribution from the equation.
There are specific constraints on "free" in that context. For example *GPL also generally forbids providing functionality without making the source available.
There will always be people putting information out there with 0 expectation of any kind of return. What will happen is there will be far fewer websites enforcing return, as they'll be automatically deranked for that enforcement.
I am tech blogging for free, and my expected return is people reading my content, recognition, attribution, social media attention and people discussing my content.
I have zero interest in providing free labor for LLM companies with no human actually reading my words. I don't think I'm alone in that stance.
Similarly, I used to help people on forums for free. My reward was getting respect of my peers, the feeling of helping another human being and sometimes them being grateful, rare side job opportunities thanks to people finding my specialist posts. That was fun, being anonymous question-answering bot for AI to scrape is not.
> recognition, attribution, social media attention
Expectation of these in particular makes your blogging a product.
> with no human actually reading my words
But it - or at least the idea - is being ultimately read by humans, as long as some article is in the top results for relevance to some LLM prompts. It just may be summarized, or particular parts extracted first, or reworded for clarity (like I may ask for an "eli5" if I encounter something that's interesting but I find concepts going over my head), or otherwise preprocessed in order to fulfill the prompt parameters. All actions which the very human users may have to do manually anyway in order to efficiently consume your content (or give up and move on if they can't be bothered), is now automatically done by LLM agents.
The number of people who have the privilege to publish content for free for others (or spend their free time contributing to open source) is minimal. And now they will feel like they're being disrespected and exploited by LLM's and its users who think of the authors not as people but as just as the source of content. You can be sure most of the content will be gone because nobody wants to be just a content cow for LLM's.
Also have the same experience with regular Google search now (and other engines, too). I think part of the problem is the degradation of Google's product, but also the endless amount of absolute shit that humanity is vomiting onto the Internet now. The commercialisation of (almost) the entire Internet during the mid 2000s has wrecked it all.
Now AI searches that: search, pull various pages to examine real contents, continue searching etc, then summarise/answer is realistically the only way to filter through all of said bullshit.
AI searches help with the clickbait problem as well, since even "reputable" news outlets are engaging in that fuckery now.
It's either; we use AI to sift through the dead carcass of our old friend, or we enforce rules for a clean Internet - which I can definitely not see happening.
Yeah I really don't believe that this is really the case, especially when we had a report recently saying clicks are down.
It has become shockingly common to see people sharing a screenshot of an AI response as evidence to back up their argument. I was once busy with something so I asked my partner if he could look up something for me, he confidently shared a screenshot of an AI response from Google. It of course was completely wrong and I had to do my own searching anyways (annoyingly needing to scroll past and ignore the AI response that kept trying to tell me the same wrong information).
We have to remember that google is incentivized to keep people on Google more. Their ability to just summarize stuff instead of getting people off of google as quickly as possible is a gold mine for them, of course they are going to push it as hard as possible.
If I search "why is rum healthy", ai overview tells this, which is... Laughable: While not a health drink, moderate consumption of rum may offer some potential benefits, primarily due to its antioxidant content and potential to improve circulation and reduce stress. Darker rums, in particular, contain higher levels of antioxidants from the aging process in wooden barrels, which can help neutralize free radicals. Additionally, rum may have a relaxing effect and can be a social lubricant, potentially reducing stress and promoting relaxation when consumed in moderation.
So I was curious, in normal google search (with the AI summary) I put in "why is rum healthy" I got this, and then it listed a bunch of benefits: "Rum, when consumed in moderation, may offer a few potential health benefits. These include possible positive effects on heart health due to its potential to increase HDL (good) cholesterol, and the presence of antioxidants in darker rums, which may be beneficial."
But if I just simply remove the "why" it clearly states "Rum is an alcoholic beverage that does not have any significant health benefits."
Man I love so much that we are pushing this technology that is clearly just "garbage in, garbage out".
Side Note: totally now going to tell my doctor I have been drinking more rum next time we do blood work if my good cholesterol is still a bit low. I am sure he is going to be thrilled. I wonder if I could buy rum with my HSA if I had a screenshot of this response... (\s if really necessary)
Well, both of those are arguments humans have repeated quite a bit. The first one is pretty weak (and you can guess who funded the “science” behind it) but it is believed by many.
Asking AI to tell reality from fiction is a bit much when the humans it gets its info from can’t, but this is at least not ridiculous.
> Asking AI to tell reality from fiction is a bit much when the humans it gets its info from can’t, but this is at least not ridiculous.
I agree with that, but the problem is that it is being positioned as a reliable source of information. And is being treated as such. Google's disclaimer "AI responses may include mistakes. Learn more" only shows up if you click the button to show more of the response, is smaller text, a light gray, and clearly overshadowed by the button with lights rotating around it to do a deep dive.
The problem is just how easy it is to "lead on" one of these models. By simply stating a search like "why is rum healthy" implies that I already think it is healthy so of course it leads into that but that is why this is so broken. But "is rum healthy" actually provides a more factual answer:
> Rum is an alcoholic beverage that does not have any significant health benefits. While some studies have suggested potential benefits, such as improved blood circulation and reduced risk of heart disease, these findings are often based on limited evidence and have not been widely accepted by the medical community.
> the problem is that it is being positioned as a reliable source of information. And is being treated as such
That's because of SEO. Top results are assumed reliable, because there is - currently - no other way to ascertain reliability in an efficient and scalable way, and the top results are sites that have optimized their content and keywords to be in the top results.
Rum is a social lubricant and it is healthy to be relaxed in social settings... quite frankly as long as they dont say health benefit increase proportionally with consumption, I see no problem saying alcohol is healthy. Or, like, why is it allowed, why was it created, why do we drink it all the time with no visible issue (beyond abuse).
If I remove why and search is rum healthy, I get this (only the beginning): While not a health drink, moderate rum consumption may offer some potential benefits. These include supporting cardiovascular health, aiding digestion, and potentially offering some antioxidant properties. However, excessive alcohol consumption can lead to serious health problems.
A sibling comment speculated that Google's response is expected because it's just a summary of bad search results. To test this, I ran a fresh query against o3 with the Web Search option enabled. The results:
> Rum is a tasty way to get ethanol into your body, but it is not a health food. Any modest upsides you may have read about (a bump in “good” HDL cholesterol, slight blood-thinning, a bit of antioxidant pickup from barrel aging) are the same things you’d get from any spirit, and recent large studies show those benefits are either tiny or disappear once you control for lifestyle. Meanwhile the well-documented downsides—cancer risk, liver disease, high blood pressure, weight gain, addiction, injuries—kick in from the first drink.
It goes on to give an extensive debunking of all the nonsense health benefits in the Google summary. It looks like it saw the same search results but was able to course correct and view those results with a more skeptical eye. Full output: https://chatgpt.com/share/6894885d-51dc-800e-ab8c-55af0a5bb2...
This post is about Google, so it's fair to focus on their search product, but I'd be careful to generalize to other LLMs. There are huge differences in quality between different models.
Are the results any different from the AI overview? Cause I assume Google search is not trying to "fix" your question and just returns relevant results. And being an "AI overview" and not an "AI answer" it should just overview the results.
The whole "Why is (false statement)?" Is an old issue and I'm not entirely convinced the Gemini lite model doing AI overviews is who we hope to fix that.
I mean I understand this response. You’re specifically asking it why rum is healthy so it’s giving you the few positive things that are potentially healthy about consuming rum.
A bit unrelated, but I did this because a few days ago I made a web search like "spirit free rum" or something similar, and in the tab slightly below the top with similar results and answers there was the first being "why rum is healthy" which said similar things to what I got with Ai overview. This is in italian though so ymmv in other languages
Why is that laughable? Rum isn't a health drink, but if you were looking for information to support the case that it has some health benefits (which is literally the search term)...seems like a reasonable answer. What did you expect? A moralistic essay on how alcohol is bad?
people make stuff up and post online. You will find made up shit with or without AI with that kind of query. So yes, it's reasonable that AI exposes you to the real Internet, and it's doing, at worst, as good a job as search engines.
A lot of people are desperate for AI to lecture to them from a position of authority, consider it broken when it doesn't, and start praying to it when it does.
edit: AI doesn't even have a corrupting, disgusting physical body, of course it should be recommending clean diets and clean spirits!
People are also more likely to click into web content that helps them learn more — such as an in-depth review, an original post, a unique perspective or a thoughtful first-person analysis
So... not the blog spam that was previously prioritized by Google Search? It's almost as if SEO had some downsides they are only just now discovering.
1) Clicking on search results doesn't bring $ to Google and takes users off their site. Surely they're thinking of ways to address this. Ads?
2) Having to click off to another site to learn more is really a deficiency in the AI summary. I'd expect Google would rather you to go into AI mode where they control the experience and have more opportunities to monetize. Ads?
We are in the "early uber" and "early airbnb" days ... enjoy it while it's great!
No SEO is required for quality content, as by definition qualitative content contains the words and terms that make it qualitative. The problem is the low quality spam that's deliberately SEO'd to masquerade as quality content.
Mandatory AI summaries have come to Google, and they gleefully showcase hallucinations while confidently insisting on their truth. I feel about them the same way I felt about mandatory G+ logins when all I wanted to do was access my damn YouTube account: I hate them. Intensely.
But why listen to a third party when you can hear it from the horses mouth.
With AI Overviews and more recently AI Mode, people are able to ask questions they could never ask before. And the response has been tremendous: Our data shows people are happier with the experience and are searching more than ever as they discover what Search can do now.
As a long-time AI+HCI person, I have mixed feelings about "AI", but just last night I was remarking to colleagues/friends that even I have mostly stopped clicking through from Google searches. The "AI" summary now usually plagiarizes a good enough answer.
I'm sure Google knows this, and also knows that that many of these "AI" answers wouldn't pass any prior standard of copyright fair use.
I suspect Google were kinda "forced" into it by the sudden popularity of OpenAI-Microsoft (who have fewer ethical qualms) and the desire to keep feeding their gazillion-dollar machine rather than have it wither and become a has-been.
"If we don't do it, everyone else will anyway, and we'll be less evil with that power than those guys." Usually that's just a convenient selfish rationalization, but this time it might actually be true.
Still, Google is currently ripping off and screwing over the Web, in a way that they still knew was wrong as recently as a few years ago, pre-ChatGPT.
Google News was definitely doing this level of "summary" before ChatGPT. I'm don't think OpenAI-MS have fewer ethical qualms, just Google had more recent memories of the negative consequences.
Okay now compare it back to when Google search used to be good in like 2006 before it would serve you barely tangentially related crap and before being optimized to prioritize spam garbage and that could have been written by a monkey on a typewriter with a finite amount of time.
I can't say for others, but this is what I do since Google integrated AI to the search results. For 80% of the time, I'd just type a question and read the AI summary and stop going further. For the other 20% or so when I believe deep diving is important, I'd scroll through results in the first page and click on a few of them to find out the "facts" myself.
The latter is what I used to do before AI summary was a thing, so I would logically assume that it should reduce the clicks to individual sites?
More like you can find the Google ai summary and then the first and only page of results are SEO optimized ai summaries by content mills. Original content is long lost and taken offline due to lack of revenue
I imagine it depends on the kind of search people are making.
if I just need a basic fact or specific detail from an article, and being wrong has no real world consequences, I'll probably just gamble it and take the AI's word for it most of the time. Otherwise I'm going to double check with an article/credible source
if anything, I think aimode from google has made it easier to find direct sources for what I need. A lot of the times, I am using AI for "tip of the tongue" type searches. I'll list a lot of information related to what I am trying to find, and the aimode does a great job of hunting it down for me
ultimately though, I do think some old aspects of google search are dying - some good, some bad.
Pros: don't fee the need to sift through blog spam, I don't need to scroll past paid search results, I can avoid the BS part of an article where someone goes through their entire life story before the actual content (I'm talking things like cooking website)
Cons: Google is definitely going to add ads to this tool at some point, some indie creators on the internet will have a harder time getting their name out.
my key takeaway from all this is that people will only stop at your site if they think your site will have something to offer that the AI can't offer. and this isn't new. people have been steeling blog content and turning into videos for ever. people will steel paid tutorials and release the content for free on a personal site. people will basically take content from site-X and repost in a more consumable format on site-Y. and this kind of theft is so obvious and no one liked seeing the same thing reposted a 1000 times. I think this long term is a win
Right but most of those clicks go to sites Google has deals with. You only get traffic that pays if you're big enough to sue Google for stealing your shit.
True. Googled "how to auto switch dark theme bootstrap". AI says in "versions 5.3 and newer, you can leverage the data-bs-theme attribute and JavaScript." and shows `data-bs-theme="auto"`.
This attribute exists, but this value comes from a bootstrap plugin that you have to install separately. It generated quite a few clicks and high-quality searches from me.
You’re raising a very important concern — the slow disappearance of human-curated knowledge niches. While AI can summarize the obvious and the popular, it struggles to preserve the quirky, community-driven, and idiosyncratic corners of the early internet. Forums and specialty sites were full of experiments, debates, and lived experiences — not just canonical facts.
If we don’t actively archive, incentivize, or reimagine those spaces, AI-generated content may become a sterile echo chamber of what’s “most likely,” not what’s most interesting. The risk isn’t that knowledge disappears — it’s that flavor, context, and dissent do.
I always wonder about this: What happens to their ads business? Also, what's the incentive for websites to provide data to Google if they’re not getting the incoming clicks? The generative approach seems to disincentivize both, right?
That, I think, is a very interesting question!
I guess Google found itself in a situation where they had to jump on the AI bandwagon and add AI features to their search. Summaries existed before for certain topics, now they pop up always.
In the long run they probably need to integrate ads into these responses or they find another way to monetize the combination of "knowing about user intent" and "present matching answers".
I’m sure it is, Google, but can you at least give me a warning before you pull out and finish on my back this time when you release the next even more invasive portion of your AI assault on unwitting, unconsenting users? Thanks.
I’m sick of having to feel violated every step I take on the Web these days.
Gaslight, gaslight, gaslight. That's all Google does now. Their AI tools gaslight us, and their human leadership gaslights us. It's all truly sad, considering what a major breakthrough Google Search was in the early days.
Use DuckDuckGo, use Kagi, use virtually anything OTHER than Google.
And also the overt political bias and censorship (it's a problem over at Youtube, which is owned by Google, they might be separate in theory but it's possible it's not)
Dumb AI is one thing. Not autocompleting "Donald Trump assassination attempt" (or any number of other things) is a choice
Liz Reid staked her career on "AI" working in search. Lo and behold, a blog post by her confirms that "AI" is working.
I've seen many outrageously wrong summaries that were contradicted sometimes by articles on the first page of regular search. Are people happy with the slop? Maybe, but I could see people getting bored by it very quickly. There already is a healthy comment backlash against ChatGPT-generated voice over narratives in YouTube videos.
"As a search company, we care passionately — perhaps more than any other company — about the health of the web ecosystem." Yeah, right, complete and utter bullshit. This really reads like the crisis PR of a parasitical ad tech business.
Based on my own usage patterns I don't think this is too implausible. When I do use an LLM chatbot for a "search", I'm almost always gathering initial information to use in one or more traditional search queries to find authoritative sources.
It does kind of contradict my own assumption that most people just take what the chatbot says as gospel and don't look any deeper, but I also generally think it's a bad idea to assume most people are stupid. So maybe there's a bit of a contradiction there.
I think it is pretty safe to assume that at least one third of Googles users will take what the chatbot says as gospel and will not look any deeper, just as users previously took the first search result as the verbatim truth.
Thank you. That's different than my use of AI summaries, which is "ignore them". I know that I want definitive info, so I look for deeper info than a summary immediately.
But I also share your assumption about "most people".
"Google says" is editorializing. When others submit content from my blog they do not say "Evan Boehs says," they just take my title. Sometimes this leads to odd titles which I'm sure you notice from time to time, like product annoucements might be "Filibuster 3" and you're like "well what is Filibuster" but such is policy.
I can believe this. A lot of my google search usage now is something like:
> "what is the type of wrench called for getting up into tight spaces"
> AI search gives me an overview of wrench types (I was looking for "basin wrench")
> new search "basin wrench amazon"
> new search "basin wrench lowes"
> maps.google.com "lowes"
Notably, the information I was looking for was general knowledge. The only people "losing out" here are people running SEO-spammish websites that themselves (at this point) are basically hosting LLM-generated answers for me to find. These websites don't really need to exist now. I'm happy to funnel 100% of my traffic to websites that are representing real companies offering real services/info (ship me a wrench, sell me a wrench, show me a video on how to use the wrench, etc).
> The only people "losing out" here are people running SEO-spammish websites that themselves (at this point) are basically hosting LLM-generated answers for me to find.
And anybody who creates original content and wishes -- not just to be paid for that content -- but for people to actually see that content and engage with it. IOW the very people who fed the LLM revolution.
> The only people "losing out" here are people running SEO-spammish websites that themselves (at this point) are basically hosting webpages containing LLM-generated answers for me to find.
Agreed. The web will be better off for everyone if these sites die out. Google is what brought these into existence in the first place, so I find it funny Google is now going to be one of the ones helping to kill them. Almost like they accidentally realized SEO got out of control so they have to fix their mistake.
At one point these SEO pages were in fact providing a real service, and you could view them as a sort of "manual", prototypical, distributed form of AI. Millions of people trying to understand what information was valuable and host webpages to satisfy the demand for that info, and get rewarded for doing so. It obviously went too far, but at one point, it did make sense to allow these websites to proliferate. I know without AI, I probably just would have clicked on the first link that said "types of wrenches" and read a little bit. I probably would have gotten my answer, it just wouldn't have been quite as laser-targeted to my exact question.
True, the early days these sites were genuinely helpful. The monetization model was a little different though which is what I think kept them useful. You'd use the content just to drive traffic, which would result in ad clicks on your banner ads, etc.
Then "content marketing" took over, and the content itself was now also used to sell a product or service, sort of an early form of influencer marketing and that is when I think it all started to go down hill. We stopped seeing the more in depth content which actually taught something, and more surface level keywords that were just used to drive you to their product/service.
OTOH, the early web was also full of niche forums, most viewable without an account and indexable, of about any topic you could imagine where you could interact with knowledgeable folks in that niche. Google would have been more helpful to users by surfacing more of those forums vs. the blogs.
Those forums are IMO the real loss here. Communities have moved into discord, or another closed platform that doesn't appear on the web, and many that require accounts or even invitations to just view read only.
"Live by the SEORD, die by the SEORD"
Or was it, "Live by the AdWord..."
Hard disagree. I put a great deal of work into my website, putting hard-earned information on the internet for the first time. Now Google reaps all the value I create without as much as a "thank you".
Unfortunately the new victims in the system of LLMs and the way they distribute knowledge is the specialized content website.
Now an LLM just knows all the content you painstakingly gathered on your site. (It could also be, and is likely that it was also collected from other hard to find sites across the internet).
The original web killed off the value of a certain kind of knowledge (encyclopedias, etc.) and LLMs will do the same.
There are plenty of places to place the blame, but this is a function of any LLM, and a funcamental way LLMs work, not just a problem created and profited from by Google- for example the open-weight models, where no-one is actually profiting directly.
> There are plenty of places to place the blame, but this is a function of any LLM, and a funcamental way LLMs work, not just a problem created and profited from by Google- for example the open-weight models, where no-one is actually profiting directly.
First time learning that scraping and training on data that they have often been explicitly disallowed to obtain for free or for that purpose by the rights holders is "fundamental to how LLMs work". If not, then there is no reason those who gathered the information wouldn't stand to profit by selling LLMs this data.
We killed forums and now were killing specialty websites. Soon we will get right back to pre internet levels of knowledge. Ai will be able to find all the stuff that was easily available in encyclopedias and text books and none of the stuff that made 90s-2000 internet great.
The Internet had plenty of specialty knowledge before "Gathering that knowledge, SEOing it, and stuffing ads all over it" became a viable business model, and it will have plenty of specialty knowledge after this era is over.
Specialty knowledge can also be a guy reviewing air purifiers with obsessive care. It can be someone writing detailed travel guides or in my case an essential resource for immigrants. There were a lot of good websites, and if you could not find them, blame Google for letting its search engine go to the dogs.
Gathering that knowledge is work, and if anyone should capture that value, it's the people doing the work. Seeing bug tech slurp it all up, insert inself in the middle and capture all the value is heartbreaking.
I hate that AI is destroying the economics of the independent web, and people cheer for it because they landed on Forbes one too many times. It's insulting to all the people who did their job right, and still get their work slurped up by an uncaring corporation with no way to stop them.
We will get the web that we deserve.
I wonder how long it will take AI agents to automatically make accounts and submit credit card info to scrape data that is paywalled
I don't think there is any reason they couldn't do that
That was already dying if not dead thanks to YouTube and short form video being the preferred way for people to make content and I guess consume
> There are plenty of places to place the blame, but this is a function of any LLM, and a funcamental way LLMs work, not just a problem created and profited from by Google- for example the open-weight models, where no-one is actually profiting directly.
One problem with monopolies is their massive multiplicative effects on otherwise manageable problems.
>The original web killed off the value of a certain kind of knowledge (encyclopedias, etc.) and LLMs will do the same.
I wonder how many people will decide to just stop sharing technical knowledge because of that, and how much we will lose because of it.
Maybe (I can't emphasize enough that I say this with a high degree of skepticism) they'll come up with a way of doing cited/credited/paid work where the LLM will include who "taught" it something.
Eg: instead of writing a blog post, you'll submit a knowledge article to an AI provider and that'll go into the AI's training set and it'll know "you" told it. And maybe (even more skeptical) pay you for it.
Again: highest degree of skepticism, but at the same time, that's the only way I could see people continuing to write content that teaches anything.
This will only happen as an absolute necessity, for example if the EU threatens massive fines or the US antitrust people grow some teeth.
Even then, you'll get malicious compliance. The best case scenario would be a bit like Spotify: everyone getting fractions of a penny.
Agreed. I'm particularly pessimistic about how this will play out.
Unfortunately, existence is a thankless job. Just gotta enjoy the ride.
You put your information on a public website for free, and are surprised that people use it as they see fit?
If I gave my surplus eggs for free, I'd be annoyed if they ended up at the Best Western's buffet. These eggs were meant for my neighbours.
There is a difference between working for free for your community and working for free for a trillion dollar company's investors. Doubly so when you strip consent or attribution from the equation.
Then you should make sure you are giving the eggs to your neighbors and not to Best Western.
Just because something is on a public website accessible for free doesn't mean it is then public domain. Sharing is not necessarily giving.
(Though unfortunately in the Wild West Web of today it seems it does, practically speaking.)
Not always creators puts the content for free free, they expect views, gratitude, and/or ad revenue..
It's not free if there's any kind of expected return.
Attribution would be one thing. See ooen-source and free software licenses.
There are specific constraints on "free" in that context. For example *GPL also generally forbids providing functionality without making the source available.
If the return is also free, yes. A smile or a thank you is free to give.
Don't be surprised if there is no more free information in the next 10 years.
There will always be people putting information out there with 0 expectation of any kind of return. What will happen is there will be far fewer websites enforcing return, as they'll be automatically deranked for that enforcement.
I am tech blogging for free, and my expected return is people reading my content, recognition, attribution, social media attention and people discussing my content.
I have zero interest in providing free labor for LLM companies with no human actually reading my words. I don't think I'm alone in that stance.
Similarly, I used to help people on forums for free. My reward was getting respect of my peers, the feeling of helping another human being and sometimes them being grateful, rare side job opportunities thanks to people finding my specialist posts. That was fun, being anonymous question-answering bot for AI to scrape is not.
> recognition, attribution, social media attention
Expectation of these in particular makes your blogging a product.
> with no human actually reading my words
But it - or at least the idea - is being ultimately read by humans, as long as some article is in the top results for relevance to some LLM prompts. It just may be summarized, or particular parts extracted first, or reworded for clarity (like I may ask for an "eli5" if I encounter something that's interesting but I find concepts going over my head), or otherwise preprocessed in order to fulfill the prompt parameters. All actions which the very human users may have to do manually anyway in order to efficiently consume your content (or give up and move on if they can't be bothered), is now automatically done by LLM agents.
The number of people who have the privilege to publish content for free for others (or spend their free time contributing to open source) is minimal. And now they will feel like they're being disrespected and exploited by LLM's and its users who think of the authors not as people but as just as the source of content. You can be sure most of the content will be gone because nobody wants to be just a content cow for LLM's.
Also have the same experience with regular Google search now (and other engines, too). I think part of the problem is the degradation of Google's product, but also the endless amount of absolute shit that humanity is vomiting onto the Internet now. The commercialisation of (almost) the entire Internet during the mid 2000s has wrecked it all.
Now AI searches that: search, pull various pages to examine real contents, continue searching etc, then summarise/answer is realistically the only way to filter through all of said bullshit.
AI searches help with the clickbait problem as well, since even "reputable" news outlets are engaging in that fuckery now.
It's either; we use AI to sift through the dead carcass of our old friend, or we enforce rules for a clean Internet - which I can definitely not see happening.
The LLM probably wouldn’t know the answer to your initial question without the SEO spam sites that they were trained on.
Yeah I really don't believe that this is really the case, especially when we had a report recently saying clicks are down.
It has become shockingly common to see people sharing a screenshot of an AI response as evidence to back up their argument. I was once busy with something so I asked my partner if he could look up something for me, he confidently shared a screenshot of an AI response from Google. It of course was completely wrong and I had to do my own searching anyways (annoyingly needing to scroll past and ignore the AI response that kept trying to tell me the same wrong information).
We have to remember that google is incentivized to keep people on Google more. Their ability to just summarize stuff instead of getting people off of google as quickly as possible is a gold mine for them, of course they are going to push it as hard as possible.
> especially when we had a report recently saying clicks are down.
Isn't that expected from "higher quality clicks"?
If I search "why is rum healthy", ai overview tells this, which is... Laughable: While not a health drink, moderate consumption of rum may offer some potential benefits, primarily due to its antioxidant content and potential to improve circulation and reduce stress. Darker rums, in particular, contain higher levels of antioxidants from the aging process in wooden barrels, which can help neutralize free radicals. Additionally, rum may have a relaxing effect and can be a social lubricant, potentially reducing stress and promoting relaxation when consumed in moderation.
So I was curious, in normal google search (with the AI summary) I put in "why is rum healthy" I got this, and then it listed a bunch of benefits: "Rum, when consumed in moderation, may offer a few potential health benefits. These include possible positive effects on heart health due to its potential to increase HDL (good) cholesterol, and the presence of antioxidants in darker rums, which may be beneficial."
But if I just simply remove the "why" it clearly states "Rum is an alcoholic beverage that does not have any significant health benefits."
Man I love so much that we are pushing this technology that is clearly just "garbage in, garbage out".
Side Note: totally now going to tell my doctor I have been drinking more rum next time we do blood work if my good cholesterol is still a bit low. I am sure he is going to be thrilled. I wonder if I could buy rum with my HSA if I had a screenshot of this response... (\s if really necessary)
Well, both of those are arguments humans have repeated quite a bit. The first one is pretty weak (and you can guess who funded the “science” behind it) but it is believed by many.
Asking AI to tell reality from fiction is a bit much when the humans it gets its info from can’t, but this is at least not ridiculous.
> Asking AI to tell reality from fiction is a bit much when the humans it gets its info from can’t, but this is at least not ridiculous.
I agree with that, but the problem is that it is being positioned as a reliable source of information. And is being treated as such. Google's disclaimer "AI responses may include mistakes. Learn more" only shows up if you click the button to show more of the response, is smaller text, a light gray, and clearly overshadowed by the button with lights rotating around it to do a deep dive.
The problem is just how easy it is to "lead on" one of these models. By simply stating a search like "why is rum healthy" implies that I already think it is healthy so of course it leads into that but that is why this is so broken. But "is rum healthy" actually provides a more factual answer:
> Rum is an alcoholic beverage that does not have any significant health benefits. While some studies have suggested potential benefits, such as improved blood circulation and reduced risk of heart disease, these findings are often based on limited evidence and have not been widely accepted by the medical community.
> the problem is that it is being positioned as a reliable source of information. And is being treated as such
That's because of SEO. Top results are assumed reliable, because there is - currently - no other way to ascertain reliability in an efficient and scalable way, and the top results are sites that have optimized their content and keywords to be in the top results.
Rum is a social lubricant and it is healthy to be relaxed in social settings... quite frankly as long as they dont say health benefit increase proportionally with consumption, I see no problem saying alcohol is healthy. Or, like, why is it allowed, why was it created, why do we drink it all the time with no visible issue (beyond abuse).
If I remove why and search is rum healthy, I get this (only the beginning): While not a health drink, moderate rum consumption may offer some potential benefits. These include supporting cardiovascular health, aiding digestion, and potentially offering some antioxidant properties. However, excessive alcohol consumption can lead to serious health problems.
Google's "AI Overview" feature is particularly bad.
o3 gives the following response:
> Short answer: it isn’t. Rum isn’t “healthy” any more than other booze.
It goes on to give a bulleted list on why it's not healthy. Full output: https://chatgpt.com/share/6894871f-be9c-800e-8cc9-9e5ec2d5d6...
A sibling comment speculated that Google's response is expected because it's just a summary of bad search results. To test this, I ran a fresh query against o3 with the Web Search option enabled. The results:
> Rum is a tasty way to get ethanol into your body, but it is not a health food. Any modest upsides you may have read about (a bump in “good” HDL cholesterol, slight blood-thinning, a bit of antioxidant pickup from barrel aging) are the same things you’d get from any spirit, and recent large studies show those benefits are either tiny or disappear once you control for lifestyle. Meanwhile the well-documented downsides—cancer risk, liver disease, high blood pressure, weight gain, addiction, injuries—kick in from the first drink.
It goes on to give an extensive debunking of all the nonsense health benefits in the Google summary. It looks like it saw the same search results but was able to course correct and view those results with a more skeptical eye. Full output: https://chatgpt.com/share/6894885d-51dc-800e-ab8c-55af0a5bb2...
This post is about Google, so it's fair to focus on their search product, but I'd be careful to generalize to other LLMs. There are huge differences in quality between different models.
Are the results any different from the AI overview? Cause I assume Google search is not trying to "fix" your question and just returns relevant results. And being an "AI overview" and not an "AI answer" it should just overview the results.
The whole "Why is (false statement)?" Is an old issue and I'm not entirely convinced the Gemini lite model doing AI overviews is who we hope to fix that.
TBF it's merely providing a summary of the top (SEO) results. The quarrel/ridicule should be with/of the sites providing that content.
I mean I understand this response. You’re specifically asking it why rum is healthy so it’s giving you the few positive things that are potentially healthy about consuming rum.
A bit unrelated, but I did this because a few days ago I made a web search like "spirit free rum" or something similar, and in the tab slightly below the top with similar results and answers there was the first being "why rum is healthy" which said similar things to what I got with Ai overview. This is in italian though so ymmv in other languages
Why is that laughable? Rum isn't a health drink, but if you were looking for information to support the case that it has some health benefits (which is literally the search term)...seems like a reasonable answer. What did you expect? A moralistic essay on how alcohol is bad?
there is no antioxidant health benefits from rum. how is making stuff up reasonable.
people make stuff up and post online. You will find made up shit with or without AI with that kind of query. So yes, it's reasonable that AI exposes you to the real Internet, and it's doing, at worst, as good a job as search engines.
> reasonable that AI exposes you to the real Internet,
Any response will be 'reasonable' by that standard.
The AI overview is just summarizing the top results. You’d find the exact same information if it wasn’t there and just clicked the search results.
> The AI overview is just summarizing the top results.
Nope
nope what, here's a tip: next time add "truth only" in your query, it works!
A lot of people are desperate for AI to lecture to them from a position of authority, consider it broken when it doesn't, and start praying to it when it does.
edit: AI doesn't even have a corrupting, disgusting physical body, of course it should be recommending clean diets and clean spirits!
From the article:
People are also more likely to click into web content that helps them learn more — such as an in-depth review, an original post, a unique perspective or a thoughtful first-person analysis
So... not the blog spam that was previously prioritized by Google Search? It's almost as if SEO had some downsides they are only just now discovering.
Google search never prioritized blog spam, spammers optimized their spam around Google's ranking algorithm
Well:
1) Clicking on search results doesn't bring $ to Google and takes users off their site. Surely they're thinking of ways to address this. Ads?
2) Having to click off to another site to learn more is really a deficiency in the AI summary. I'd expect Google would rather you to go into AI mode where they control the experience and have more opportunities to monetize. Ads?
We are in the "early uber" and "early airbnb" days ... enjoy it while it's great!
SEO and quality content are goals that should slowly align with time, so I consider this convergence very welcome.
They converged very briefly, and have since diverged significantly. I don't expect that trend to reverse.
No SEO is required for quality content, as by definition qualitative content contains the words and terms that make it qualitative. The problem is the low quality spam that's deliberately SEO'd to masquerade as quality content.
Counterpoint from yesterday.
https://news.ycombinator.com/item?id=44798215
From that article
But why listen to a third party when you can hear it from the horses mouth.Whether you believe the the article or not, the point you posted seems orthogonal to what google is saying.
They're not claiming anything about the quality of AI summaries. They are analyzing how traffic to external sites has been affected.
The first paragraph in the article is
“Our data shows people are happier with the experience and are searching more than ever as they discover what Search can do now.”
One can also interpret this as search was such shit that the summaries are allowing users to skip that horrible user experience.
They don’t care about discoverability. It’s all ads as quickly as possible. Coming soon is ad links in summaries. That’s what they’re getting to here.
And they're searching more than ever because Google is failing to actually serve useful content that the user was looking for.
No. They’re searching more because the summaries are giving them new keywords or questions to ask.
As a long-time AI+HCI person, I have mixed feelings about "AI", but just last night I was remarking to colleagues/friends that even I have mostly stopped clicking through from Google searches. The "AI" summary now usually plagiarizes a good enough answer.
I'm sure Google knows this, and also knows that that many of these "AI" answers wouldn't pass any prior standard of copyright fair use.
I suspect Google were kinda "forced" into it by the sudden popularity of OpenAI-Microsoft (who have fewer ethical qualms) and the desire to keep feeding their gazillion-dollar machine rather than have it wither and become a has-been.
"If we don't do it, everyone else will anyway, and we'll be less evil with that power than those guys." Usually that's just a convenient selfish rationalization, but this time it might actually be true.
Still, Google is currently ripping off and screwing over the Web, in a way that they still knew was wrong as recently as a few years ago, pre-ChatGPT.
Google News was definitely doing this level of "summary" before ChatGPT. I'm don't think OpenAI-MS have fewer ethical qualms, just Google had more recent memories of the negative consequences.
Okay now compare it back to when Google search used to be good in like 2006 before it would serve you barely tangentially related crap and before being optimized to prioritize spam garbage and that could have been written by a monkey on a typewriter with a finite amount of time.
I can't say for others, but this is what I do since Google integrated AI to the search results. For 80% of the time, I'd just type a question and read the AI summary and stop going further. For the other 20% or so when I believe deep diving is important, I'd scroll through results in the first page and click on a few of them to find out the "facts" myself.
The latter is what I used to do before AI summary was a thing, so I would logically assume that it should reduce the clicks to individual sites?
More like you can find the Google ai summary and then the first and only page of results are SEO optimized ai summaries by content mills. Original content is long lost and taken offline due to lack of revenue
Anecdotally, Google search is back to #1 for me.
Having an optional digest of the SERP makes link selection easier, especially if I have only a rough idea of what I'm looking for.
I imagine it depends on the kind of search people are making.
if I just need a basic fact or specific detail from an article, and being wrong has no real world consequences, I'll probably just gamble it and take the AI's word for it most of the time. Otherwise I'm going to double check with an article/credible source
if anything, I think aimode from google has made it easier to find direct sources for what I need. A lot of the times, I am using AI for "tip of the tongue" type searches. I'll list a lot of information related to what I am trying to find, and the aimode does a great job of hunting it down for me
ultimately though, I do think some old aspects of google search are dying - some good, some bad.
Pros: don't fee the need to sift through blog spam, I don't need to scroll past paid search results, I can avoid the BS part of an article where someone goes through their entire life story before the actual content (I'm talking things like cooking website)
Cons: Google is definitely going to add ads to this tool at some point, some indie creators on the internet will have a harder time getting their name out.
my key takeaway from all this is that people will only stop at your site if they think your site will have something to offer that the AI can't offer. and this isn't new. people have been steeling blog content and turning into videos for ever. people will steel paid tutorials and release the content for free on a personal site. people will basically take content from site-X and repost in a more consumable format on site-Y. and this kind of theft is so obvious and no one liked seeing the same thing reposted a 1000 times. I think this long term is a win
Keeping in mind, turning off typo corrections would also drive more queries and higher quality clicks.
Has the AI delisted geeks4geeks? That'd be a massive improvement.
Right but most of those clicks go to sites Google has deals with. You only get traffic that pays if you're big enough to sue Google for stealing your shit.
True. Googled "how to auto switch dark theme bootstrap". AI says in "versions 5.3 and newer, you can leverage the data-bs-theme attribute and JavaScript." and shows `data-bs-theme="auto"`.
This attribute exists, but this value comes from a bootstrap plugin that you have to install separately. It generated quite a few clicks and high-quality searches from me.
Counterpoint: https://news.ycombinator.com/item?id=44142113
You’re raising a very important concern — the slow disappearance of human-curated knowledge niches. While AI can summarize the obvious and the popular, it struggles to preserve the quirky, community-driven, and idiosyncratic corners of the early internet. Forums and specialty sites were full of experiments, debates, and lived experiences — not just canonical facts.
If we don’t actively archive, incentivize, or reimagine those spaces, AI-generated content may become a sterile echo chamber of what’s “most likely,” not what’s most interesting. The risk isn’t that knowledge disappears — it’s that flavor, context, and dissent do.
> the quirky, community-driven, and idiosyncratic corners of the early internet
This will be able to come again to the fore once SEO'd spam dies off due to click starvation.
hi AI
I always wonder about this: What happens to their ads business? Also, what's the incentive for websites to provide data to Google if they’re not getting the incoming clicks? The generative approach seems to disincentivize both, right?
That, I think, is a very interesting question! I guess Google found itself in a situation where they had to jump on the AI bandwagon and add AI features to their search. Summaries existed before for certain topics, now they pop up always. In the long run they probably need to integrate ads into these responses or they find another way to monetize the combination of "knowing about user intent" and "present matching answers".
Any concrete data to support this claim by Google?
its surely giving me more clicks to find things, not sure about "higher quality". Probably generates more ad revenue
I’m sure it is, Google, but can you at least give me a warning before you pull out and finish on my back this time when you release the next even more invasive portion of your AI assault on unwitting, unconsenting users? Thanks.
I’m sick of having to feel violated every step I take on the Web these days.
You will loose 90% traffic and you will be happy [mem]
> AI in Search is driving more queries
Because the search results are crap.
> and higher quality clicks
Because the users are fooled by some results and click on them, only to lose time and curse.
Gaslight, gaslight, gaslight. That's all Google does now. Their AI tools gaslight us, and their human leadership gaslights us. It's all truly sad, considering what a major breakthrough Google Search was in the early days.
Use DuckDuckGo, use Kagi, use virtually anything OTHER than Google.
And also the overt political bias and censorship (it's a problem over at Youtube, which is owned by Google, they might be separate in theory but it's possible it's not)
Dumb AI is one thing. Not autocompleting "Donald Trump assassination attempt" (or any number of other things) is a choice
seeing how other people use search engines shakes my paradigms
seems like marketing paffle tbh
Liz Reid staked her career on "AI" working in search. Lo and behold, a blog post by her confirms that "AI" is working.
I've seen many outrageously wrong summaries that were contradicted sometimes by articles on the first page of regular search. Are people happy with the slop? Maybe, but I could see people getting bored by it very quickly. There already is a healthy comment backlash against ChatGPT-generated voice over narratives in YouTube videos.
"As a search company, we care passionately — perhaps more than any other company — about the health of the web ecosystem." Yeah, right, complete and utter bullshit. This really reads like the crisis PR of a parasitical ad tech business.
Is there any evidence that Google is telling the truth? Because this sounds like bullshit.
You think companies would do that? Just go on the internet and tell lies?!
Based on my own usage patterns I don't think this is too implausible. When I do use an LLM chatbot for a "search", I'm almost always gathering initial information to use in one or more traditional search queries to find authoritative sources.
It does kind of contradict my own assumption that most people just take what the chatbot says as gospel and don't look any deeper, but I also generally think it's a bad idea to assume most people are stupid. So maybe there's a bit of a contradiction there.
I think it is pretty safe to assume that at least one third of Googles users will take what the chatbot says as gospel and will not look any deeper, just as users previously took the first search result as the verbatim truth.
> almost always gathering initial information to use in one or more traditional search queries to find authoritative sources
For me at least with Perplexity, Grok and ChatGPT, all results come back with citations in every paragraph, so I haven't had to do that.
Thank you. That's different than my use of AI summaries, which is "ignore them". I know that I want definitive info, so I look for deeper info than a summary immediately.
But I also share your assumption about "most people".
... because they make it unavoidable and default?
Isn't editorializing the titles against the rules?
Complaining about the submission is also against the guidelines, but what are you gonna do.
You can flag the post and anyone is welcome to email us (hn@ycombinator.com) about editorialized titles or inappropriate submissions.
Was the post edited? It looks like the submission title is exactly the blog post title except with "Google Says" appended.
"Google says" is editorializing. When others submit content from my blog they do not say "Evan Boehs says," they just take my title. Sometimes this leads to odd titles which I'm sure you notice from time to time, like product annoucements might be "Filibuster 3" and you're like "well what is Filibuster" but such is policy.