They get basically everything they want (keeping it all in the tent), plus a negotiating position on search deals where they can refuse something because they can't do it now.
Quite why the judge is so concerned about the rise of AI factoring in here is beyond me. It's fundamentally an anticompetitive decision.
I basically only use Google for "take me to this web page I already know exists" queries now, and maps.
I remember the feeling when I first started using ChatGPT in late 2022, and it's the same feeling I had when Google search came out in the early 2000s. And that was like, "oh chatgpt is the new Google".
Do you check all of the sources though? Those can be hallucinated and you may not notice unless you're always checking them. Or it could have misunderstood the source.
It's easy to assume it's always accurate when it generally is. But it's not always.
The type of search you are doing probably matters a lot here as well. I use it to find documentation for software I am already moderately familiar with, so noticing the hallucinations is not that difficult. Although, hallucinations are pretty rare for this type of "find documentation for XYZ thing in ABC software" query. Plus, it usually doesn't take very long to verify the information.
I did get caught once by it mentioning something was possible that wasn't, but out of probably thousands of queries I've done at this point, that's not so bad. Saying that, I definitely don't trust LLMs in any cases where information is subjective. But when you're just talking about fact search, hallucination rates are pretty low, at least for GPT-5 Thinking (although still non-zero). That said, I have also run into a number of problems where the documentation is out-of-date, but there's not much an LLM could do about that.
This subthread is classic HN. Huge depth of replies all chiming in to state some form of the original prior: that "AI is a threat to search"...
... without even a nod to the fact that by far the best LLM-assisted search experience today is available for free at the Google prompt. And it's not even close, really. People are so set in their positions here that they've stopped even attempting to survey the market those opinions are about.
(And yes, I'm biased I guess because they pay me. But to work on firmware and not AI.)
The only people who are being homogenized or "down-graded" by Chat GPT are people who wouldn't have sought other sophisticated strategies in the first place, and those who understand that Chat GPT is a tool and understand how it works, and it's context, can utilize it efficiently with great positive effect.
Obviously Chat GPT is not perfect but it doesn't need to be perfect to be useful. For a search user, Google Search has not been effective for so long it's unbelievable people still use it. That is, if you believe search should be a helpful tool with utility and not a product made to generate maximum revenue at the cost of search experience.
Would you say that people were losing braincells using google in 2010 to look up an animal fact instead of going to a library and opening an encyclopedia?
Eventually those answers will be sufficient for most and give people no reason to move to alternatives.
Allowing them to pay to be default seems to mostly guarantee this outcome
That said their "Dive into AI" feature has cause me to use it more lately.
A recentish example, I was trying to remember which cities' buses were in Thessaloniki before they got a new batch recently. They used to rent from a company (Papadakis Bros) that would buy out of commission buses from other cities around the world and maintain the fleet. I could remember specifically that there were some BVG Busses from Berlin, and some Dutch buses, and was vaguely wondering if there were some also from Stockholm I couldn't remember.
So I searched on my iPad, which defaulted to Google (since clearly I hadn't got around to setting up a good search engine on it yet). And I get this result: https://i.imgur.com/pm512HU.jpeg
The LLM forced its way in there without me prompting (in e.g. Kagi, you opt in by ending the query with a question mark). It fundamentally misunderstands the question. It then treats me like an idiot for not understanding that Stockholm is a city in Sweden, and Thessaloniki a city in Greece. It uses its back linking functionality to help cite this great insight. And it takes up the entire page! There's not a single search result in view.
This is such a painful experience, it confirms my existing bias that since they introduced LLMs (and honestly for a couple years before that) that Google is no longer a good first place to go for information. It's more of a last resort.
Both ChatGPT and Claude have a free tier, and the ability to do searches. Here's what ChatGPT gave me: https://chatgpt.com/share/68b78eb7-d7b4-8006-81e0-ab2c548931...
A lot of casual users don't hit the free tier limits (and indeedI've not hit any limits on the free ChatGPT yet), and while they have their problems they're both far better than the Gemini powered summaries Google have been pumping out. My suggestion is that perhaps you haven't surveyed the market before suggesting that "by far the best LLM-assisted search experience today is available for free at the Google prompt".
Assuming we're talking about the AI generated blurbs at the top of search results, there are loads of problems. For one they frequently don't load at all. For another search is an awkward place for them to be. I interact with search differently than with a chat interface where you're embedding a query in a kind of conversational context such that both your query and the answer are rich in contextual meaning. With search I'm typically more fact finding and in a fight against Google's page rank optimizations to try and break through to get my information I need. In a search context AI prompts don't benefit from context rich prompts and aren't able to give context-rich answers and kind of give generic background that isn't necessarily what I asked for. To really benefit from the search prompts I would have to be using the search bar in a prompt way, which would likely degrade the search results. And generally this hybrid interaction is not very natural or easy to optimize, and we all know nobody is asking for it, it's just bolted on to neutralize the temptation to leave search behind in favor of an LLM chat.
And though less important, material design as applied to Google web sites in the browser is not good design, it's ugly and the wrong way to have a prompt interaction. This is also the case for Gemini from a web browser. Meanwhile GPT and Claude are a bit more comfortable with information density and are better visual and interactive experiences because of it.
I agree this is annoying but other than that I really can't follow your argument: You're comparing a keyword-like "prompt" given to Google's LLM to a well-phrased question given to ChatGPT and are surprised the former doesn't produce the same results?
So like a lot of the internet? I don’t really understand this idea that LLMs have to be right 100% of the time to be useful. Very little of the web currently meets that standard and society uses it every day.
But as it stands, it's a terrible user experience. It's ugly, the page remains incredibly busy and distracting, and it is wrong far more often than ChatGPT (presumably because of inference cost at that scale).
It might be good enough to slow the bleeding and keep less demanding users on SERP, but it is not good enough to compete for new users.
I'm pretty sure they meant LLMs in general, not just ChatGPT. They all straight up lie to very similar degrees, no contest there.
> The only people who are being homogenized or "down-graded" by Chat GPT are people who wouldn't have sought other sophisticated strategies in the first place, and those who understand that Chat GPT is a tool and understand how it works, and it's context, can utilize it efficiently with great positive effect.
I know for a fact that this isn't true. I have a friend who was really smart, probably used to have an IQ of 120 and he would agree with all of this. But a few of us are noticing that he's essentially being lobotomized by LLMs and we've been trying to warn him but he just doesn't see it, he's under the impression that "he's using LLMs efficiently with great positive effect".
In reality his intellectual capabilities (which I used to really respect) have withered and he gets strangely argumentative about really basic concepts that he's absolutely wrong about. It seems like he won't accept it as true until an LLM says so. We used to laugh at those people together because this could never happen to us, so don't think that it can never happen to you.
Word of advice for anyone reading this: If multiple people in your life suddenly start warning you that your LLM interactions seem to be becoming a problem for one reason or another, make the best possible effort to hear them out and take them seriously. I know it probably sounds absurd from your point of view, but that's simply a flaw in our own perception of ourselves, we don't see ourselves objectively, we don't realize when we've changed.
https://freedium.cfd/https://vinithavn.medium.com/from-multi...
At its core, attention operates through three fundamental components — queries, keys, and values — that work together with attention scores to create a flexible, context-aware vector representation.
Query (Q): The query is a vector that represents the current token for which the model wants to compute attention.
Key (K): Keys are vectors that represent the elements in the context against which the query is compared, to determine the relevance.
Attention Scores: These are computed using Query and Key vectors to determine the amount of attention to be paid to each context token.
Value (V): Values are the vectors that represent the actual contextual information. After calculating the attention scores using Query and Key vectors, these scores are applied against Value vectors to get the final context vectorWhile that's concerning, my own experience in seeking information using this approach has been positive: it provides a fast, fully customised answer that easily outweighs the mistakes it makes. This flattens the learning curve on a new subject and with that saved time I am able to confirm important details to weed out the mistakes/hallucinations. Whereas with Googling I'd be reading technical documentation, blog posts and whatever else I could find, and -crucially- I'd still need to be confirming the important details because that step was never optional. Another plus is that I'm now not subjected to low quality ai-generated blog spam when seeking information.
I foresee Google search losing relevance rapidly, chatbots are the path of least resistance and "good enough" for most tasks, but I also am aware that Google's surveillance-based data collection will continue to be fruitful for them regardless if I use Google search or not.
A documentation for a specific product I expect to be mostly right, but maybe miss the required detail.
Some blog, by some author I haven't heard about I trust less.
Some third party sites I give some trust, some less.
AI is a mixed bag, while always implying authority on the subject. (While becoming submissive when corrected)
There are buses in Stockholm, and buses in Thessoloniki, and buses manufactured in Sweden, and buses previously used in Stockholm that are now in operation in Thessoloniki. And one LLM took one path through the question, answering it correctly and completely. And the other took a different one[1]. As it happened your (poorly phrased) intended question was answered by one and not the other.
If I ask the same question with a more careful phrasing that (I think!) matches what you wanted to know: "Where did buses used in Thessoloniki come from originally?"
...I get correct and clear answers from both. But the Google result also has the Wikipedia page for the transit operator and its own web page immediately to the right.
Again, cherry picking notwithstanding I think in general the integrated experience of "I need an AI to help me with this problem" works much better at google.com, it just does.
[1] It's worth pointing out that the result actually told you that your question didn't make sense, and why. I suspect you think this was a bug since the other LLM guessed instead, but it smells like a feature to me.
Which is exactly my point. A bunch of people doing that to conform with the shibboleth identity of the phone in their pocket and then posting strong opinions about the product they don't (or at least claim not to) use is an echo chamber and not a discussion. You only get the upvotes in these threads if you conform.
HN is supposed to be better than that.
Someone here has religion, and it’s not me. I don’t use Google search because it’s a terrible product and we finally have other options. As for AI, there are dozens of options, and it does not take many examples to see how bad Ai Overview is. Gemini 2.5 Pro, however, is in my tool belt.
I know. But you're posting confidently (along with a ton of other people) in a subthread about Google search anyway, making statements about its behavior which you straight up admit to be unqualified to make. And I'm calling out the disconnect, because someone has to.
No one in an echo chamber thinks they're in an echo chamber. This is 100% an echo chamber.
I suffered through Google search's decline for the last 15 years along with everyone else. I land on it often enough still to see that the trend is not changing.
And if it is true... it is not a common experience and would have external factors contributing to this behavior.
Additionally, using IQ to qualify someone's intelligence is a signal, so I'll just not go into it deeper since we will disagree as I find your anecdote juvenile, straight up exaggeration, or a complete lie to serve your opinion.
Plausibly this could happen if you had the ego of a 16 year old or were socially disabled, and it would be alleviated over time through experience. I'm not trying to be rude but you sound like a Tiktok conspirator and I'm old enough and experienced enough to smell bullshit.
I have a really good idea of his "abilities" and we've been in contact regularly all these years which is why I can pinpoint when this all started. I even talked to a few of our common friends about it because I'm worried about his well-being and they agreed that there's been a pretty dramatic shift in his abilities and demeanour.
> using IQ to qualify someone's intelligence is a signal
You're way off base. Believe me I know that IQ as such isn't a great way of measuring and especially comparing intelligence between cultures, but I was comparing him against himself.
In other words, he used to be really smart and had above-average critical thinking skills. Now he's seemingly incapable of critically engaging with new technical ideas or reason about their implications. He just refuses to engage with anything that is mentally demanding.
> I'm not trying to be rude [...] and I'm old enough
You should probably put more effort into the "trying not to" part. Your entire response is "straight up" a rude dismissal accusing me of being a "juvenile" "TikTok" liar at every step, "lol".