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

Simply explained: How does GPT work?

confusedbit.dev

Post reply on HN
Print view View original

311–320 of 392 posts

Re: Simply explained: How does GPT work?

#311

Post by simonh »

iliane5 wrote: post #231
tines wrote: post #145

Earlier quoted context omitted.

> If just predicting the next token can produce similar or better results than the almighty human intelligence on some tasks But it's not better than almighty human intelligence, it _is_ human intelligence, because it was trained on a mass of some of the best human intelligence in all recorded history (I say this because the good stuff like Aristotle got preserved while the garbage disappeared (this was true until th…

> But it's not better than almighty human intelligence, it _is_ human intelligence, because it was trained on a mass of some of the best human intelligence in all recorded history Sure, I was saying "better" in the sense that if for X task, it can do better than Y% of humans. > since we hand-fed it the answers, it falls a little flat for me We didn't really hand-fed it any answers though did we? If you put a human in…

I think the human would make a lot of the same fundamental errors LLMs make, for similar reasons. The level to which LLMs seem to understand the world is highly superficial because it is entirely linguistic. Also human written texts about the world and human affairs miss out huge swathes of contextual information that we safely assume actual humans have. LLMs don’t have any of that, which is why they fall flat on their faces in so many ways.

Re: Simply explained: How does GPT work?

#312

Post by simonh »

lordnacho wrote: post #220
kmod wrote: post #200

Earlier quoted context omitted.

Here's an example that I think garners more agreement that properties of a limit ("really understanding") don't necessarily mean that any path towards that limit has the properties of the limit. I think there's a lot of room for disagreement about whether this is a factually-accurate analogy and I'm not trying to argue either way on that, just trying to answer your question about how one might make these sorts of arg…

I think what this points towards is that we care about the internal mechanism. If we prod it externally and it gives the wrong answer, then the internal mechanism is definitely wrong. But if we get the right answers and then open it up and find the internals are still wrong, it's still wrong. This illuminates a contradiction: the walks like a duck thing is incompatible with the internals being a duck. If you see a cr…

> I think what this points towards is that we care about the internal mechanism.

The mechanism is what makes a system interesting.

In software this is why we develop libraries of algorithms and code we can reuse and compose into new solutions. The programmer is providing the intellectual flexibility, while the code is the set of capabilities. It’s why this is a superior approach, compared to building a single monolithic mass of procedural code from scratch in a single variable scope for every program we write.

Solutions matter because it’s not just about what a system can do now, it’s about what it can learn or be adapted to do next.

Re: Simply explained: How does GPT work?

#313

Post by simonh »

iliane5 wrote: post #224
simonh wrote: post #183

Earlier quoted context omitted.

I think it’s undeniable that LLMs encode knowledge, but the way they do so and what their answers imply, compared to what the same answer from a human would imply, are completely different. For example if a human explains the process for solving a mathematical problem, we know that person knows how to solve that problem. That’s not necessarily true of an LLM. They can give such explanations because they have been tra…

Well it's like birds and airplanes. Do airplanes "fly" in the same sense that birds do? Of course not, birds flap their wings and airplanes need to be built, fueled and flown by humans. You could argue that the way birds fly is "more natural" or superior in some ways but I've yet to see a bird fly Mach 3. If you replace the analogy with humans and LLMs, LLMs won't ever reason or understand things in the same way we d…

I think the issue is there are good reasons to think LLMs architected and trained the way they are now can never approach human reasoning capability. That’s because the corpus of human written material is simply grossly inadequate to communicate or encode the knowledge necessary for that.

Our written material assumes huge swathes of contextual knowledge, real world experience, and human lived experience that LLMs don’t and can’t have. At least architected and trained as they are now.

Thats on top of the crippling inability LLMs have to generalise an ability to perform a task from the ability to generate a description of how to do the task. Plus many other similar limitations that would be inexplicable if displayed by a human.

Of course LLMs aren’t the final word in AI development. I think they’re a vitally important step towards general AI, and we’ll get there eventually as we develop ever more capable architectures.

Re: Simply explained: How does GPT work?

#314

Post by simonh »

opportune wrote: post #261
iliane5 wrote: post #136

Earlier quoted context omitted.

What I find really entertaining is the "just predicting the next token" argument. If just predicting the next token can produce similar or better results than the almighty human intelligence on some tasks, then maybe there's a bit of hubris in how smart we think we actually are.

I don’t think there’s anything making it impossible for actual intelligence to arise from a task as simple as “predicting the next token (to model human thought/speech/writing)” because with enough compute resources, smart AI implementations, and training that task basically would be optimized by becoming a general intelligence. But it’s clear based on current implementations that once you work backwards from the kno…

The problem is such an approach is limited by the content of the training texts. As I mentioned elsewhere, our written texts assume huge swathes of contextual and experiential information and knowledge that LLMs don’t have. It’s possible some of it might be inferred from the texts, but not all of it by a long shot.

If somehow you could generate a training text encoding a complete and thorough understanding of the physical world, human psychology and sociology, and reasoning then that might get you quite far. But the existing it even near future human textual corpus isn’t really that. Even then I still think you’d hit the limitations of the LLM cognitive architecture pretty hard.

Re: Simply explained: How does GPT work?

#315

Post by roflyear »

JieJie wrote: post #269
bulkprotocol wrote: post #184

Earlier quoted context omitted.

You blew up a 30 year friendship over an...analogy?

I didn't! Someone else did it to me. I was trying desperately not to. (edit: This is the kind of stuff I think my friends are watching and being informed by [0] as it was what they are posting in our common areas.) [0]: https://youtu.be/ro130m-f_yk

Why does it upset you if they disagree with you on this?

Re: Simply explained: How does GPT work?

#316

Post by i-use-nixos-btw »

root_axis wrote: post #242
i-use-nixos-btw wrote: post #130

I’d be interested in hearing from anyone who takes the Chinese Room scenario seriously, or at least can see how it applies to any of this. I cannot see that it matters if a computer understands something. If it quacks like a duck and walks like a duck, and your only need is for it to quack and walk like a duck, then it doesn’t matter if it’s actually a duck or not for all intents and purposes. It only matters if you…

> if it quacks like a duck and walks like a duck, and your only need is for it to quack and walk like a duck, then it doesn’t matter if it’s actually a duck or not for all intents and purposes. This isn't true. If a plane flies like a bird and you only need it for flying it doesn't then follow that a plane is a bird "for all intents and purposes".

The emphasis on “for all intents and purposes” is important. If you specify requirements you specify them for your intent and for your purpose.

Requiring that something fly and that something be a plane are two different things with only minor overlap. If all you require is something that flies, then a dragonfly matches your requirements exactly as much as an apache helicopter does.

Re: Simply explained: How does GPT work?

#317

Post by iliane5 »

simonh wrote: post #311
iliane5 wrote: post #231

Earlier quoted context omitted.

> But it's not better than almighty human intelligence, it _is_ human intelligence, because it was trained on a mass of some of the best human intelligence in all recorded history Sure, I was saying "better" in the sense that if for X task, it can do better than Y% of humans. > since we hand-fed it the answers, it falls a little flat for me We didn't really hand-fed it any answers though did we? If you put a human in…

I think the human would make a lot of the same fundamental errors LLMs make, for similar reasons. The level to which LLMs seem to understand the world is highly superficial because it is entirely linguistic. Also human written texts about the world and human affairs miss out huge swathes of contextual information that we safely assume actual humans have. LLMs don’t have any of that, which is why they fall flat on the…

Absolutely. What’s fascinating is that they’re getting such good understanding of many things through just text. Multimodal models that can process text, images, sounds, video, etc. are gonna be very interesting for that very reason

Re: Simply explained: How does GPT work?

#318

Post by i-use-nixos-btw »

missingrib wrote: post #238
i-use-nixos-btw wrote: post #170

Earlier quoted context omitted.

Yes, in this instance I understand failings of today (though copilot has a much better hit rate, and at the moment it’s a great augmentation to coding if you treat it like an enthusiastic intern). My question is about the future. The argument goes that a machine can never understand Chinese, even if it is capable of interpreting Chinese and responding to or acting on the input perfectly every time. My reply is that,…

> My reply is that, if it acts as if it understands Chinese in every situation, then there’s no measurable way of distinguishing it from understanding. I'm not sure if you understood the argument. The argument isn't asserting that there is a measurable way of distinguishing it, it's actually claiming that regardless of how well it seems like it understands Chinese, it doesn't actually understand Chinese. It's about i…

You are entirely correct: I do not understand the argument - or, rather, the relevance of it in any applied situation.

In effect, my argument is that in order for you to require it to understand something, you require it to understand that thing for a reason. If it acts like it understands that thing under all probing, then your requirement is satisfied - the question about whether it truly understands the thing is moot, because it fulfils the requirement.

Re: Simply explained: How does GPT work?

#319

Post by iliane5 »

simonh wrote: post #313
iliane5 wrote: post #224

Earlier quoted context omitted.

Well it's like birds and airplanes. Do airplanes "fly" in the same sense that birds do? Of course not, birds flap their wings and airplanes need to be built, fueled and flown by humans. You could argue that the way birds fly is "more natural" or superior in some ways but I've yet to see a bird fly Mach 3. If you replace the analogy with humans and LLMs, LLMs won't ever reason or understand things in the same way we d…

I think the issue is there are good reasons to think LLMs architected and trained the way they are now can never approach human reasoning capability. That’s because the corpus of human written material is simply grossly inadequate to communicate or encode the knowledge necessary for that. Our written material assumes huge swathes of contextual knowledge, real world experience, and human lived experience that LLMs don…

> LLMs architected and trained the way they are now can never approach human reasoning capability

Not sure if you’ve played with GPT-4 but honestly it’s getting there. If you take the bar exam, ChatGPT was in the bottom 10% of participants, GPT-4 is in the top 90%.

It obviously isn’t the ultimate test of reasoning/intelligence but I think we would agree that a human who’s in the top 90% is likely to be pretty smart.

> Of course LLMs aren’t the final word in AI development

Couldn’t agree more. AGI will come from plugging a few of these systems together.

Re: Simply explained: How does GPT work?

#320

Post by iliane5 »

majormajor wrote: post #258
iliane5 wrote: post #136

Earlier quoted context omitted.

What I find really entertaining is the "just predicting the next token" argument. If just predicting the next token can produce similar or better results than the almighty human intelligence on some tasks, then maybe there's a bit of hubris in how smart we think we actually are.

We've known for a long time that computers can do calculations far, far, far faster than us. We continue to figure out new ways to make those calculations do more complicated things faster than humans. What is intelligence beyond calculation is an ancient question, but not the one I'm most interested in at the moment, re: today's tools. I'm curious right now about if there's meaning to other people in human creation…

I think as soon as text2video gets really good (like midjourney level), there’s gonna be so much AI generated content that unless it’s all extremely good, human made content will be something people search specifically for.

As for curation, I think the success of TikTok proves that you don’t need that much data to pretty preceding pinpoint what someone wants to watch (or what will get them to spend the most time on the app at least).

Post reply on HN