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

"Hallucinating" AIs sound creative, but let's not celebrate being wrong

thereader.mitpress.mit.edu

Post reply on HN
Print view View original

161–170 of 196 posts

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#161

Post by pdonis »

gaganyaan wrote: post #147
pdonis wrote: post #58

Earlier quoted context omitted.

> All these criticisms of hallucinations are missing the point. Not if people need to be reminded that, as you say, LLMs are not designed to give reliable answers. Many, many people appear to believe that they are, since they rely on the answers in all kinds of contexts. For example, in past HN threads on LLMs, I have seen people say that they rely on LLM-generated code.

I rely on code written by AI and verified by me, particularly when the code is finicky or tedious but verification is simple. LLMs are more reliable than you're trying to assert, but blindly copy/pasting AI-generated code is a recipe for pain, just the same as blindly copy/pasting human-generated code from SO.

> I rely on code written by AI and verified by me

The "verified by me" part is crucial, though, yes?

I'm talking about people, whom I have seen posting here on HN in other threads on this topic, who leave out the "verify" step and just rely on the LLM code as it comes.

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#162

Post by mjr00 »

gaganyaan wrote: post #150
mjr00 wrote: post #34

Earlier quoted context omitted.

> This is an odd article. To me, it seems like the "creative" arts are an ideal arena for AI. After all, there's no such thing as "wrong" art. Yeah but who wants to consume art purely generated by AI (that is, not human-created with AI support)? Most art sites have had blanket bans, or at least required tagging, on ai-generated art because people hate it so much. Or to put it another way: why are you in the comment s…

> who wants to consume art purely generated by AI Most people won't care one bit. Some artists care strongly, hence the bans on artist-focused sites. When it's about as good or better than humans, people will consume it. As for why people are still on HN, TBH a lot of these comments could be easily written by ChatGPT. What makes it worth it is seeing comments from people in the know, which is something ChatGPT can't…

> When it's about as good or better than humans, people will consume it.

How are you measuring better? AI art is already better than most humans in a lot of ways: better shading, better coloring, even better anatomy outside of hands and feet (which has been mostly solved with controlnet and related extensions). Yet nobody wants to look at it.

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#163

Post by gaganyaan »

mjr00 wrote: post #162
gaganyaan wrote: post #150

Earlier quoted context omitted.

> who wants to consume art purely generated by AI Most people won't care one bit. Some artists care strongly, hence the bans on artist-focused sites. When it's about as good or better than humans, people will consume it. As for why people are still on HN, TBH a lot of these comments could be easily written by ChatGPT. What makes it worth it is seeing comments from people in the know, which is something ChatGPT can't…

> When it's about as good or better than humans, people will consume it. How are you measuring better? AI art is already better than most humans in a lot of ways: better shading, better coloring, even better anatomy outside of hands and feet (which has been mostly solved with controlnet and related extensions). Yet nobody wants to look at it.

> Yet nobody wants to look at it.

That's simply false. Lots of people are enjoying it. Especially porn of course, but people are enjoying plenty of regular AI art, right now. "Better" isn't objective, more people will enjoy more AI art as it gets better at generating art that they like.

What makes you think otherwise?

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#164

Post by gaganyaan »

pdonis wrote: post #157
gaganyaan wrote: post #148

Earlier quoted context omitted.

> LLMs are just generating text. I really can't stand these low-effort takes. It's like saying "Human brains are just meat". You can summarize anything badly, but it's not helpful to the conversation. Whatever quibbles you have aside, LLMs are generating useful text, today. At some point you're just pointlessly privileging meat over silicon.

> It's like saying "Human brains are just meat". No, it isn't, because human brains have rich semantic connections to the rest of the world. LLMs do not. So "LLMs are just generating text" is a justified description of their limitations, in a way that "human brains are just meat" is not.

> LLMs do not

Prove it

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#165

Post by pests »

ryukoposting wrote: post #99

Spot-on article, for the most part. I firmly believe that the use of an LLM as a fact-finding tool represents a misunderstanding of the technology's abilities and intent. If you're using an LLM like it's a search engine, then you should just use a search engine instead. LLMs are conversational devices. Search engines are an index of human knowledge. Right tool, right job. I'm fine with the term "hallucination," thoug…

The chat features are a fine tuning of the LLM tho. Not all LLMs are the conversational chat style.

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#166

Post by gaganyaan »

pdonis wrote: post #161
gaganyaan wrote: post #147

Earlier quoted context omitted.

I rely on code written by AI and verified by me, particularly when the code is finicky or tedious but verification is simple. LLMs are more reliable than you're trying to assert, but blindly copy/pasting AI-generated code is a recipe for pain, just the same as blindly copy/pasting human-generated code from SO.

> I rely on code written by AI and verified by me The "verified by me" part is crucial, though, yes? I'm talking about people, whom I have seen posting here on HN in other threads on this topic, who leave out the "verify" step and just rely on the LLM code as it comes.

Yes, verification (or trust) is crucial. Those posts may have elided the verification process they did, though. Unless we look at a specific post it'd be hard to discuss it.

To be more specific, I trust GPT-4 to handle tasks like "Rewrite this C++ struct into Rust". I still verify that it's correct as part of a normal code review process, but I trust GPT-4 in that scenario at least as much as I would a junior dev.

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#167

Post by ckyngcybra »

famouswaffles wrote: post #86
pdonis wrote: post #61

Earlier quoted context omitted.

> Language models (the computation) "care" about making "correct" predictions Predictions of what? LLMs aren't predicting anything. They are, as the GP says, generating text that appears to be fluent and coherent (as long as we don't inquire into actual semantic relationships with anything else) in the context of the prompt.

They are quite literally trained to predict a distribution that could supply the next token given previous context. This is their primary objective.

[flagged]

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#168

Post by teaearlgraycold »

thfuran wrote: post #156
teaearlgraycold wrote: post #98

Earlier quoted context omitted.

Here's a bit from an episode of This American Life that I think is relevant, cut into a post on my blog: https://danangell.com/blog/posts/gpt-understands/ > The idea is that back when GPT-4 was being trained for it to really consistently get the next word correct, to do that reliably, it had to do more than just bullshit. It had to do more than guess based on patterns. To get the next word right, it had to truly unde…

Does an analog circuit understand differential equations?

I think it's not really interesting to say so, but by the same definition it would be fair to describe an analog circuit that models some system as understanding the system in some capacity.

It's more interesting when the thing doing the understanding is a massive system that has both mastered English and other domains and then incidentally modeled some systems English can describe. An analog circuit that models artillery trajectories "understands" one kinematic equation. But that's kind of obvious and mundane. Understanding just one equation barely registers as profound and without that understanding existing in a larger context the word barely seems applicable.

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#169

Post by ckyngcybra »

happytiger wrote: post #123
yladiz wrote: post #118

Earlier quoted context omitted.

Please link to published papers that prove what you’re positing.

To tack on, I would love to read more research on this topic if there is relevant research to read. I have not read much compelling evidence though I have seen a lot of researchers conjecturing and some inappropriately reaching (and anthropomorphising the living hell out of ai in the process sometimes). There’s a lot to learn and I don’t think anyone can keep up with all the papers coming out so let’s keep things pos…

[flagged]

Re: "Hallucinating" AIs sound creative, but let's not celebrate being wrong

#170

Post by jshen »

pixl97 wrote: post #57
jshen wrote: post #44

Earlier quoted context omitted.

> "The LLM AI technology generation is optimized to be fluently conversational and not to be factually correct all the time." I always find this point a bit odd because humans aren't "optimized" to be correct either. > "LLMs have no understanding of the underlying reality" I struggle with this one because I see both sides of it. I was making a prompt the other day and gave a CSV file as an input and told the LLM if c…

>I always find this point a bit odd because humans aren't "optimized" to be correct either. It depends which subsystem. The older the system in the body the less likely it is to have a high error rate, otherwise we'd die from cancer at a much higher rate or injure ourselves far more often. Of course this also depends on the definition of 'correct', if the system never changed we'd never evolve. >> "LLMs have no under…

I'm talking about all of the cognitive science which seems to point to the fact that we often care more about fitting in with our group that being correct. Group cohesion seems to have been selected for.

Edit: there is also loads of evidence that our brains are wired to make incorrect decisions in many cases. Loss aversion is one example.

Post reply on HN