Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT Search

openai.com

Post reply on HN
Print view View original

871–880 of 1001 posts

Re: ChatGPT Search

#871

Post by carlosjobim »

generalizations wrote: post #547
hn_throwaway_99 wrote: post #525

Earlier quoted context omitted.

> I'm also cautiously optimistic though. We'll get there, but it's gonna be a bit shakey for a minute or two. But I don't understand how all of these AI results (note I haven't used Kagi so I don't know if it's different) don't fundamentally and irretrievably break the economics of the web. The "old deal" if you will is that many publishers would put stuff out on the web for free, but then with the hope that they cou…

> Why would anyone ever publish stuff on the web for free unless it was just a hobby That's exactly what the old deal was, and it's what made the old web so good. If every paid or ad-funded site died tomorrow, the web would be pretty much healed.

Only middle-class and rich people could participate in "the old deal" Internet made by and for hobbyists. I think people forget this. It was not so democratized and open for everyone – you first had to afford a computer.

If you're a member of a yacht club, you can probably expect other members to help you out with repairs while you help them. But when a club has half the world population as members, those arrangements don't work anymore.

Re: ChatGPT Search

#872

Post by jumping_frog »

pocketarc wrote: post #552
hn_throwaway_99 wrote: post #525

Earlier quoted context omitted.

> I'm also cautiously optimistic though. We'll get there, but it's gonna be a bit shakey for a minute or two. But I don't understand how all of these AI results (note I haven't used Kagi so I don't know if it's different) don't fundamentally and irretrievably break the economics of the web. The "old deal" if you will is that many publishers would put stuff out on the web for free, but then with the hope that they cou…

The internet was great before the great monetization of it, had tons of information provided for free with no ads. After ads, it will still have tons of information. Stack Overflows will still exist, Wikipedias, corporate blogs that serve just to boost the company, people making courses and other educational content, personal blogs (countless of which make their way here), all of those will continue to exist. Ad-driv…

Nobody will visit stackoverflow because AI through its reasoning and back and forth with users will have solved the problems. This process creates training data for future AIs of that particular company unavailable to any other

Re: ChatGPT Search

#873

Post by LegionMammal978 »

DSingularity wrote: post #137
jsheard wrote: post #115

Earlier quoted context omitted.

Given how many of those SEO spam sites are themselves generated by ChatGPT now, OpenAI can simply back-reference their own logs to find out which sites are probably SEO spam while everyone else is left guessing. That's vertical integration!

I’m sure they will be more subtle than that otherwise it will get circumvented. I’m sure they will/are tackling this at the model level. Train them to both generate good completions while also embedding text with good performance at separating generated and human text.

Personally, I'm a pessimist on this front. People assert that a model-in-training can effortlessly sift out the real data from mountains of LLM spam. But then people also assert that AI detectors do not work and can never work, since LLM output is simply too good, and any watermarking can be broken up by a light paraphrasing step. It doesn't make much sense to have it both ways.

I can only await companies' attempts to publish enough junk to create an 'alternative truth' for new LLMs to believe in. The worst part is, it might even work.

Re: ChatGPT Search

#874

Post by HarHarVeryFunny »

marcus_holmes wrote: post #651
freediver wrote: post #241

Been thinking about this a lot [1]. Will this fundamentally change how people find and access information? How do you create an experience so compelling that it replaces the current paradigm? The future promised in Star Trek and even Apple's Knowledge Navigator [2] from 1987 still feels distant. In those visions, users simply asked questions and received reliable answers - nobody had to fact-check the answers ever. C…

I agree that this is the core question, but I'd put it as: Who Gets To Decide What Is True? With a search paradigm this wasn't an issue as much, because the answers were presented as "here's a bunch of websites that appear to deal with the question you asked". It was then up to the reader to decide which of those sites they wanted to visit, and therefore which viewpoints they got to see. With an LLM answering the que…

> I agree that this is the core question, but I'd put it as: Who Gets To Decide What Is True?

Well, outside of matters of stark factuality (what time does the library close?, what did MSFT close at?), many things people may be "searching for" (i.e. trying to find information about) are more in the realm of informed opinion and summary where there is no right or wrong, just a bunch of viewpoints, some probably better informed than others.

I think this is the value of traditional search where the result is a link to a specific page/source whose trustworthiness or degree of authority you can then judge (e.g. based on known reputation).

For AI generated/summarized "search results", such as those from "ChatGPT Search" (awkward name - bit like a spork), the trustworthiness or degree of authority is only as good as the AI (not person) that generated it, and given today's state of technology where the "AI" is just an LLM (prone to hallucination, limited reasoning, etc), this is obviously a bit of an issue...

Even in the future, when presumably human level AGI will have made LLMs obsolete, I think it'll still be useful to differentiate search from RAG AGI search/chat, since it'll still be useful to know the source. At that point the specific AGI might be regarded as a specific person, with it's own areas of expertise and biases.

The name "ChatGPT Search" is very awkward - presumably they are trying to position this as a potential Google competitor and revenue generator (when the inevitable advertisements come), but at the end of the day it's just RAG.

Re: ChatGPT Search

#875

Post by karmonhardan »

veesahni wrote: post #650
greenavocado wrote: post #483

Earlier quoted context omitted.

Has google completely stopped working for anyone else? I can still search things, i get results but, they're an ordered list of popular places the engine is directing me to. Some kind of filtering is occurring on nearly every search i make that's making the results feel entirely useless. Image search stopped working sometime ago and now it just runs an AI filter on whatever image you search for, tells you there's a m…

it seems google switched to some kind of semantic search instead of word based search. It doesn't search for what your words are, but instead what your words kindof sortof mean. Which means you always get 'average' results instead of precise results. This may make it better for the non-tech folks who search for things in unclear language, and likely make it worse for those who search with precision (i.e. much of the…

People have been throwing up the, "You're just too technical," canard for years, and it's just not true. Google search does not find what anyone is looking for; it finds what it wants to sell, in a way that makes it more likely that a purchase will be made. Full stop.

Re: ChatGPT Search

#876

Post by jumping_frog »

mewpmewp2 wrote: post #405
psychoslave wrote: post #326

Earlier quoted context omitted.

What do you mean with "when it starts"? To my mind it's obvious all LLM are heavily biased to a point it's ridiculous, all the more with the confident tone they are trained to take. I have no doubt Chinese LLM will prise the party as much as American ones will sing the gospel of neoliberal capitalism.

We could have many different LLMs trained to be biased in different ways and have some form of bias checking tester, like ground news - then everyone can get their bias and live in their echo chamber.

The reason many echo chambers exist is to keep the centre of mass or barycenter from moving too fast.

Re: ChatGPT Search

#877

Post by netdevnet »

freediver wrote: post #241

Been thinking about this a lot [1]. Will this fundamentally change how people find and access information? How do you create an experience so compelling that it replaces the current paradigm? The future promised in Star Trek and even Apple's Knowledge Navigator [2] from 1987 still feels distant. In those visions, users simply asked questions and received reliable answers - nobody had to fact-check the answers ever. C…

> Path forward requires solving the core challenge: actually surfacing the content people want to see, not what intermiediaries want them to see

But this will never happen with mainstream search imo. It is not a technical problem but a human one. As long as there is a human in control of what gets surfaced, it is only a matter of time until you revert to tampered search. Humans are not robots. They have emotions and can be swayed with or without their awareness. And this is a form of power for the swayer as much as oil or water are.

The idea that you can have an AI system provide factual and reliable answers to human centric questions is as real as Star Trek itself.

You will never remove the human factor from AI

Your hope might be that a technical solution is found for a human problem but that is unlikely.

Re: ChatGPT Search

#878

Post by dahart »

newsclues wrote: post #765
lazerlapin wrote: post #743

Earlier quoted context omitted.

> A better example might be "is dietary saturated fat a major factor for heart disease in Western countries?". The current government publications (which answer "yes") for this are probably wrong based on recent research. The government cannot be relied upon as a source of truth for this. I know it was just an example, but actually no, the role of dietary saturated fat as a factor for heart disease remains very much…

https://pmc.ncbi.nlm.nih.gov/articles/PMC9794145/ Remarkable how easy it is to cling to propaganda

I wasn’t aware of the new debate over saturated fats, I’m curious and will be watching this evolve with interest.

That said, for me this publication has red flags right from the start. Complaining about difficulty of changing everyone’s minds is a political and non-academic persuasion tactic that does not convince me. Calling it “resistance” and “bias” is a bullshit framing that makes me less likely to trust Teicholz. Of course there is resistance to 50 years of publication and research, and there should be. There’s a lot of bias towards the earth being round, and a lot of resistance to the idea that it’s flat, right? If I repeat the claim that the earth is round, is that “propaganda”? It would indeed take time and effort to change everyone’s minds about that.

Multiple times she references “>20” papers that back up her claims. Except 5 of her references in this paper are her own. And she has around 10 on this subject. So is she claiming this “new consensus” is based on what she herself and maybe one or two other people believe? If 50% of the evidence for consensus is her own papers, then I doubt there’s any consensus at all. It’s funny to claim there’s consensus at the same time she complains that it’s difficult to change the consensus. Even 20 independent scientific papers not authored by Teicholz is practically nothing in the big picture. It will take many more papers and much more time, and the evidence needs to be overwhelming, clear, obvious, and true.

She might be right! But Nina Teicholz is a journalist, not a scientist. She does have a PhD, but her publications don’t appear to be scientific research, and most look like opinion pieces.

Out of curiosity, if saturated fats aren’t the culprit, what is? Looks like she does have one paper questioning sugar, so is she claiming sugar is the real cause? What if it’s the combination of sugar and saturated fats? Does that make her right or wrong?

Re: ChatGPT Search

#879

Post by mistermann »

AYBABTME wrote: post #767
lazerlapin wrote: post #743

Earlier quoted context omitted.

> A better example might be "is dietary saturated fat a major factor for heart disease in Western countries?". The current government publications (which answer "yes") for this are probably wrong based on recent research. The government cannot be relied upon as a source of truth for this. I know it was just an example, but actually no, the role of dietary saturated fat as a factor for heart disease remains very much…

You can undo 50 years of research easily, by this process which we call science.

Science is[1] impressive but does it have a method for resolving the numerous conflicting "truths" in this thread?

[1] except when it isn't, of course

Re: ChatGPT Search

#880

Post by dx034 »

As a website owner, I have an incentive to provide access to the Google crawler because of the visitors they send me. They've started to show more content directly, but a fair amount of people still click through to websites.

It looks to me like ChatGPT Search hides links much more. Is there any incentive for website owners to allow access to ChatGPT?

Post reply on HN