Live data from Hacker News

Generative AI could make search harder to trust

wired.com

101–110 of 255 posts

Re: Generative AI could make search harder to trust

#101

Earlier quoted context omitted.

Yes. I love black text on white background. A rare find these days. Browsing today is like: “You ask for a spaghetti recipe and the page tell you the whole history of civilization.”

Thats specific to recipes because they can’t be copyrighted

I had a look and definitely learned something today so #til.

Also, note to self to collect my favourite recipes in markdown files from now on.

Re: Generative AI could make search harder to trust

#102
post #72

Earlier quoted context omitted.

My understanding is that google went from a more traditional database style which supported such queries, to a newer "n-gram" index with a layer of semantic similarity. Notably, you can no longer put a sentence in quotes to only find pages that contain that exact phrase. Also, the order of words matters more now than it used to (where the old search engines treated a space as AND, so order was irrelevant outside of q…

https://www.google.com/search?q=%22you+can+no+longer+put+a+s...

Hah, perhaps I should edit that to say "reliably."

Re: Generative AI could make search harder to trust

#103

I wonder if there will be a human information/knowledge equivalent of low-background steel (pre-WWII/nukes). Data from before a certain point won't be 'contaminated' with LLM stuff, but it'll be everywhere after that. https://en.wikipedia.org/wiki/Low-background_steel

People are vastly everestimating how unique this problem of hallucinations is. It seems to me it relies mostly on discounting just how much we've already had to deal with this same problem in humans over the millenia. The problem of proliferation of bad information might be getting worse, but this isn't native to generative AI. The entire informational ecosystem has to deal with this. GPTs compound the issue, but as…

humans can only produce semi-convincing bullshit at a limited rate

with AI this limit is all but removed

all the human generated bullshit ever created will soon be dwarfed by what AI can vomit out in an hour

Re: Generative AI could make search harder to trust

#104
post #24

Earlier quoted context omitted.

> The temptation to use these technologies all the time is too strong for even self preservation of your own (writers) voice. I don't know about that. I have played with ChatGPT/Copilot/etc enough to know what they're capable of doing. But the thing is, I enjoy programming. I enjoy breaking down a problem and solving it with code. I enjoy crafting elegant code. So I don't use AI even though I'm fully aware it could s…

> So I don't use AI even though I'm fully aware it could save me hours on projects. Enjoy this luxury while it lasts. Based on what I have seen in performance review committees for software developers, your peers who drive results faster than you do because they use AI will be rewarded more and will be more likely to survive rounds of layoffs when they inevitably happen.

That's fine. I genuinely wouldn't want to continue working in an industry that worked like that anyway, so I'd just quit and keep on programming with my own projects. So that luxury will last as long as I want it to.

Re: Generative AI could make search harder to trust

#105

More amusing and frightening is when people search about themselves and turn up AI generated crap. Googling yourself was always a lucky grab bag, with the possibility of long-forgotten embarrassments being dragged up. But at least you'd have to face facts. Now I hear of people discovering they're in prison, married to random people they've never met, or are actually already dead. What is this going to do to recon on…

> Now I hear of people discovering they're in prison, married to random people they've never met, or are actually already dead.

My real name is very, very common -- so this has been my reality for my entire life.

These days, I have grown to appreciate it. It's like an invisibility superpower.

Re: Generative AI could make search harder to trust

#106

Earlier quoted context omitted.

The type of sites GP is talking about are typically hosted on .edu servers, under faculty webhosting (often featuring a "/~profname/" in the url). That's a non-trivial signal.

~/name at an edu is pretty attainable. .edu domains can be had for any otherwise eligible "U.S.-based postsecondary institutions" per Educause: https://net.educause.edu/eligibility.htm > Pages at extant domains might variously be available to undergraduate or graduate students, faculty, staff, and adjuncts. Those might either directly host emulative material or be convinced or compromised into hosting content. If the…

It is not easy for a regular person to obtain access to a .edu webpage.

Re: Generative AI could make search harder to trust

#107

Earlier quoted context omitted.

People are vastly everestimating how unique this problem of hallucinations is. It seems to me it relies mostly on discounting just how much we've already had to deal with this same problem in humans over the millenia. The problem of proliferation of bad information might be getting worse, but this isn't native to generative AI. The entire informational ecosystem has to deal with this. GPTs compound the issue, but as…

humans can only produce semi-convincing bullshit at a limited rate with AI this limit is all but removed all the human generated bullshit ever created will soon be dwarfed by what AI can vomit out in an hour

Like most things in the world. The problem isn't necessarily the technology but the scale at which it is implemented.

Re: Generative AI could make search harder to trust

#108

Earlier quoted context omitted.

Those simple web 1.0 sites made by college professors are a gold-standard in my book. I always enjoy finding them in search results. Although they are becoming increasingly rare.

Can't prove it, but it seems to me like black text on white background sites from the past are poorly ranked compared to sites with "modern" layouts.

I have a website that is just a few black text on white HTML files I maintain in whatever text editor I have at hand. Loads lightning fast, and if you cannot view it, its not a web browser. Last I checked, the total site size was about 60KB.

As time goes on, even the amount of text I am putting out get trimmed down. Make the words count, don't count the words.

Re: Generative AI could make search harder to trust

#109

I wonder if there will be a human information/knowledge equivalent of low-background steel (pre-WWII/nukes). Data from before a certain point won't be 'contaminated' with LLM stuff, but it'll be everywhere after that. https://en.wikipedia.org/wiki/Low-background_steel

People are vastly everestimating how unique this problem of hallucinations is. It seems to me it relies mostly on discounting just how much we've already had to deal with this same problem in humans over the millenia. The problem of proliferation of bad information might be getting worse, but this isn't native to generative AI. The entire informational ecosystem has to deal with this. GPTs compound the issue, but as…

Uh, no. Generative AI does not have a standard of truth. It generates text according to the probabilities given what it learned from its data set. There is no systematic modelling of a world that it can test its assertions against. What are called hallucinations are a fundamental property of the approach. The hallucinations are probable -- just not true. The model has succeeded but the assertions are false. This has to be understood or the models will mess stuff up. A lot of checking is required.

Humans always assemble information according to a standard of truth. It is a big part of how humans learn. No method is perfect but the human method results in fewer routine hallucinations.

Re: Generative AI could make search harder to trust

#110

Earlier quoted context omitted.

> So I don't use AI even though I'm fully aware it could save me hours on projects. Enjoy this luxury while it lasts. Based on what I have seen in performance review committees for software developers, your peers who drive results faster than you do because they use AI will be rewarded more and will be more likely to survive rounds of layoffs when they inevitably happen.

That's fine. I genuinely wouldn't want to continue working in an industry that worked like that anyway, so I'd just quit and keep on programming with my own projects. So that luxury will last as long as I want it to.

I've come to the same conclusion. I became a programmer because in school I loved programming and saw myself loving the job, and I do.

If the job changes that drastically I'll just have to quit and find something else.

Post reply on HN