Hacker Newsnew | past | comments | ask | show | jobs | submit | PacificSpecific's commentslogin

If the hardware changes significantly and those sites don't exist in the future wouldn't that mean gemeni would degrade in quality because it has nothing to pull from?

Right, that success story is only because there was "organic" (for lack of a better term) information from an original source. What happens when all information is nth generation AI feedback with all links to the original source lost?

Edit: A question from AI/LLM ignorance- Can the source database for an LLM be one-way, in that it does not contain output from itself, or other LLMs? I can imagine a quarantined database used for specific applications that remains curated, but this seems impossible on the open internet.


> Can the source database for an LLM be one-way, in that it does not contain output from itself, or other LLMs?

I think, for public internet data, we can only be reasonably confident for information before the big release of ChatGPT.


Yes, people have likened pre-LLM Internet content to low-background steel.

If in the hypothetical future the continual learning problem gets solved, the AI could just learn from the real world instead of publications and retain that data.


That's exactly why text written before the first LLMs has a premium on it these days. So no, all major models suffer from slop in their training data.

We've all tried to ask the LLM about something outside of its training data by now.

In that situation, they give the (wrong) answer that sounds the most plausible.


> In that situation, they give the (wrong) answer that sounds the most plausible.

Not if you use web search or deep report, you should not use LLMs as knowledge bases, they are language models - they learn language not information, and are just models not replicas of the training set.


That's definitely been my experience. I work with a lot of weird code bases that have never been public facing and AI has horrible responses for that stuff.

As soon as I tried to make a todomvc it started working great but I wonder how much value that really brings to the table.

It's great for me though. I can finally make a todomvc tailored to my specific needs.


I'm not sure what sorts of weird codebases you're working with but I recently saw Claude programming well on a Lambda MOO -- weirder than that?

Once or twice, for me it's deflected rather than answer at all.

On the other hand, they've also surfaced information (later independently confirmed by myself) that I had not been able to find for years. I don't know what to make of it.


> because it has nothing to pull from?

Chat rooms produce trillions of tokens per day now, interactive tokens, where AI can poke and prod at us, and have its ideas tested in the real world (by us).


This then becomes the hardware manufacturers problem. If their new hardware fails for to many users it will no longer be purchased. If they externalize their problem solving like so many companies, they won't be able to gain market share.

This creates financial incentives to pay companies running the new version of search. Your thinking of this as a problem for these companies, when in reality it is a financial incentive.


Presumably companies will still provide manuals.

It'll be a single sheet of paper with a QR code that redirects to a canned prompt hosted at whichever LLM server paid the most to the manufacturer for their content.

If that was adequate then wouldn't there not be supplementary material?

Results vary of course. I have some very wonderful synthesizer manuals.


Yea so I’ve had an issue getting video output after boot on a new AMD R9700 Pro. None of the, albeit free, models from OpenAI/Google/Anthropic have really been helpful. I found the pro drivers myself. They never mentioned them.

Thats not to say AI is bad. It’s great in many cases. More that I’m worried about what happens when the repositories of new knowledge get hollowed out.

Also my favorite response was this gem from Sonnet:

> TL;DR: Move your monitor cable from the motherboard to the graphics card.


You could put a sim card in a tablet in that case. Might look a little funny when doing a phone call though.

I've done a couple flirty interviews and so far it hasn't come up. So take hope, it's not all bad.

I generally worked on AAA stuff but the few times I released stuff independently I ended up writing my own framework/engine.

I think the issue was when I used an engine the scope was too large and I never completed the work so I never released the game (or I released it for free because I felt it was incomplete and wasn't worth charging for)

It's great to work in a constrained environment


How is it not relevant? Celeste wasn't exactly a pinnacle of bleeding edge technology when it came out.

If I remember correctly it was a team of 2.


It was also originally created for the PICO8 platform. Which is as minimal as you can get.

Did the stamp work out?

Ideally, what should the stamp say?

It says in the OP's link. I was just wondering if book buyers respected the stamp and furthermore if the people stealing the books recognize it enough to be deterred.

I've never looked at the app store rankings before. Number 3 is a app for buying shoes? That seems quite specific.

This is a sure sign we’re nearing the shoe event horizon (https://hitchhikers.fandom.com/wiki/Shoe_Event_Horizon and audio https://m.youtube.com/watch?v=nEI19kJ5GfU )

People will always need shoes.

Yeah I still use Google for local restaurant info and geospatial stuff because it's just plain better.

I'm not sure if this is true anymore but years ago I heard from some business owners in my area that Google would call them to keep business information up to date. I imagine there's some form of automation these days (robocalls maybe?) but if that is the case it's kind of funny to think they are superior because they still do some stuff somewhat manually.


I'm gonna have to see if I can get my company to switch off openAI. Hopefully we can make a small dent and if enough of us do it, a larger dent.

Sounds like it won't really be a pain for me though based off comments on HN indicating Claude is the better product and I doubt I personally would hit any sort of token limits with the amount I use agentic coding.


Gametap was so great and really underrated at the time. I think I probably ended up introducing it to 4-5 people.

We're starting to see some gametap-esqu stuff again these days but it's like 15 years later and the quality isn't there for me. Even though my employer keeps giving me free Xbox ultimate subscriptions I never really use them. I think a big part was gametap was so frictionless, you boot up the client and start playing.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: