Live data from Hacker News

GPT-4 Designed a Programming Language

lukebechtel.com

321–330 of 338 posts

Re: GPT-4 Designed a Programming Language

#321

Earlier quoted context omitted.

You're describing over fitting to some look up table. Can't be what's happening here. Because the examples LLMs are answering are well out of bounds of the "100^2" training data. The internet is huge but it's not that huge. One can easily find chatGPT saying, doing or creating things that obviously come from a generalized model. It's actually trivial to find examples of chatGPT answering questions with responses that…

Like yes to some extent there is a mild amount of generalization in that it is not literally regurgitating the internet and it to some extent mixes text really well but I don't think that's obviously the full on generalization of understanding that humans have. These models obviously are more sample efficient at learning relationships than a literal lookup table but like I've already said: my example was obviously ex…

>I want to reemphasize another portion of my comment: it really does seem that when you step outside of the domain of the internet, the error rates rise dramatically especially when there is completely no analogous situation.

This is not surprising. A human would suffer from similar errors at a similar rate if it were exclusively fed an interpretation of reality that only consisted of text from the internet.

>These models obviously are more sample efficient at learning relationships than a literal lookup table but like I've already said: my example was obviously extreme for the purposes of illustration that sample efficiency does seem to matter. If you used 100^2 - 1 samples,

Even within the context of the internet there are enough conversational scenarios where you can have chatGPT answer things in ways that are far more generalized then "minor".

Take for example: https://www.engraved.blog/building-a-virtual-machine-inside/

Read it to the end. In the beginning you could say that the terminal emulation does exist as a similar copy in some form on the internet. But the structure that was built in the end is unique enough that it could be said nothing like it has ever existed on the internet.

Additionally you have to realize that while bash commands and results do exist in ON the internet, chatGPT cannot simply copy the logic and interactive behavior of the terminal from text. In order to do what it did (even in the beginning) it must "understand" what a shell is and it has to derive that understanding from internet text.

Re: GPT-4 Designed a Programming Language

#322

Earlier quoted context omitted.

>No, the type of errors are indicative of a complete lack of understanding. That is the point. They are errors that a thinker with an incomplete understanding would never make. They are so garbled that not even a true believer such as yourself can find a way to shoehorn a possible interpretation of correctness into them; such that you are forced to admit that the machine is in error. Otherwise you and the other belie…

> No. You're wrong. chatGPT only knows of text. It derives incomplete understanding of the world via text. It doesn't understand anything. It combines symbols according to a probabilistic algorithm and you assign meaning to it. > Why are you so stuck on this stupid concept? Because you keep replying with statements indicating you are yet to grasp it. > This is a bad analogy. I'm not claiming sentience. My claim is th…

>It doesn't understand anything. It combines symbols according to a probabilistic algorithm and you assign meaning to it.

This is what the human brain does. I'm not assigning meaning to it. I am simply saying the algorithm is isomorphic to our definition of the word "understanding". No additional meaning.

>Because you keep replying with statements indicating you are yet to grasp it.

No no. What's going on here is I'm replying with statements to help YOU understand and you are repeatedly failing.

>There is just as much evidence for sentience as understanding.

Sentience is too fuzzy of a word to discuss. We can't even fully define it. Understanding is less fuzzy and more definable thus the question and claim for "understanding" is a much more practical query.

A human can be inconsistent and even lie. It does not mean the human does not understand you. Thus because your logic is applicable to humans it is akin to saying humans don't understand you. That is why your logic is incorrect.

Re: GPT-4 Designed a Programming Language

#323

Earlier quoted context omitted.

> No. You're wrong. chatGPT only knows of text. It derives incomplete understanding of the world via text. It doesn't understand anything. It combines symbols according to a probabilistic algorithm and you assign meaning to it. > Why are you so stuck on this stupid concept? Because you keep replying with statements indicating you are yet to grasp it. > This is a bad analogy. I'm not claiming sentience. My claim is th…

>It doesn't understand anything. It combines symbols according to a probabilistic algorithm and you assign meaning to it. This is what the human brain does. I'm not assigning meaning to it. I am simply saying the algorithm is isomorphic to our definition of the word "understanding". No additional meaning. >Because you keep replying with statements indicating you are yet to grasp it. No no. What's going on here is I'm…

> This is what the human brain does.

The human brain is embodied in a human flesh and uses language to exchange models and data about the real world with other fleshy vessels. This provides a basis to assign meaning to the language. Furthermore we know that humans understand to a greater or lesser extent because we are human and have insight into the human experience of language and reality.

These machine learning algorithms lack this fundamental basis for ascribing meaning to the symbolic tokens they deal with. Furthermore we lack the common experience for inferring meaning and understanding, we have to interpret from the output whether there is meaning and understanding on the machine's end. Without access to internal experience we must always harbor some doubt but given some level of nonsensical outputs we can say with confidence that there is no indication of understanding.

> A human can be inconsistent and even lie. It does not mean the human does not understand you. Thus because your logic is applicable to humans it is akin to saying humans don't understand you. That is why your logic is incorrect.

Like everyone else, I interpret statements from humans differently than statements from machines. This is because I know that humans and machines are different, and therefore the meaning assigned to the symbols involved is also different.

Re: GPT-4 Designed a Programming Language

#324

Earlier quoted context omitted.

You say > it's intentions based on its training are to accurately predict the next character. then you say: > these agents do have desires that are analogous to our own I think its very in-human to have a single desire which is to predict the next character. That is not analogous to our desires. And the intention to predict the next character is not the intention of the chatbot, it is the intention of whoever created…

Clearly, having a single life desire and nothing else is in-human, I completely agree! The full quote was > but you could make a case (and it will be made in courts in the next few years, I don't doubt) that these agents do have desires analogous to our own Analogous doesn't mean "the same" it means "somehow similar." However I would challenge you to consider more specifically why this type of desire is different fro…

There is no "desire" in the computer. Therefore the question of whether its desire is different from our desires is meaningless, because it does not have a desire.

Computer just executes instructions. It doesn't matter to it whether its desires are fulfilled or not. Whereas humans do have desires: If we get thirsty we suffer unless we get our desire of drinking fulfilled.

Re: GPT-4 Designed a Programming Language

#325
post #225

Last year, when Blake Lemoine was fired by Google for claiming his AI was sentient, a lot of people on this website sensibly noted that the danger of these chatbots was not that they were sentient, but that average people could easily be tricked into thinking that they were. And now here we are: it seems we tech people are every bit as susceptible to this kind of thinking as the average joes we looked down on. Actual…

The interesting part of the exchange is the contrast between the surface level syntax of the exchange and the semantics below. It reminds me of teaching in a lot of ways. The level of polish on the immediately apparent surface level of the conversation is very high: reasonable tenents, good justification and ability to explain them. But then applying them directly into an example falls down a bit. The language is a b…

But it could be a good starting point for a group of humans to take, polish and maintain as a new language.

Not sure why need any new languages when we have the perfection that is Golang tho.

Re: GPT-4 Designed a Programming Language

#326

Earlier quoted context omitted.

Like yes to some extent there is a mild amount of generalization in that it is not literally regurgitating the internet and it to some extent mixes text really well but I don't think that's obviously the full on generalization of understanding that humans have. These models obviously are more sample efficient at learning relationships than a literal lookup table but like I've already said: my example was obviously ex…

>I want to reemphasize another portion of my comment: it really does seem that when you step outside of the domain of the internet, the error rates rise dramatically especially when there is completely no analogous situation. This is not surprising. A human would suffer from similar errors at a similar rate if it were exclusively fed an interpretation of reality that only consisted of text from the internet. >These m…

> This is not surprising. A human would suffer from similar errors at a similar rate if it were exclusively fed an interpretation of reality that only consisted of text from the internet.

I think this is surprising at least if the bot actually understands, especially for domains like math. It makes errors (like in adding large numbers) that shouldn't occur if it wasn't smearing together internet data. We would expect there to be many homework examples on the internet of adding relatively small numbers but less of large numbers. A large portion of what makes math interesting is that many of the structures we are interested in exist in large examples and in small examples (though not always) so if you understand the structure, it should be able to guide you pretty far. Presumably most humans (assuming they understand natural language) can read a description of addition then (with some trial and error) get it right for small cases. Then when presented with a large case would generalize easily. I don't usually guess out the output and instead internally try to generate and algorithm I follow.

> Take for example: https://www.engraved.blog/building-a-virtual-machine-inside/

When I first saw that a while back, I thought that was a more impressive example but only marginally more so than the natural language examples. Like how these models are trained under supervised learning imply that it should be able to capture relationships between text well. Like you said, there's a lot of content associating the output of a terminal with the input.

Maybe this is where I think we're miscommunicating right. I don't think even for natural language it's purely just copying text from the internet. It is capturing correlations and I would argue that simply capturing correlations doesn't imply an understanding. To some extent, it knows what the output of curl is supposed to look like and can use attention to figure out the website to then generate what an intended website is supposed to look like. Maybe the sequential nature of the commands is kind of impressive but I would argue that at least for the jokes.txt example, that particular sequence is at least probably very analogous to some tutorial on the internet. It's difficult to find since I would want to limit myself before 2021.

It can correlate the output of a shell to the input, and to some extent, the relationships between the output of a command and input are well produced and its training and suffused it with information about what terminal outputs (is this what you are referring to when you say it has to derive understanding from internet text?), but it doesn't seem to be reasoning about the terminal despite probably being trained on a lot of documentation about these commands.

Like we can imagine that this relationship is also not too difficult to capture. A lot of internet websites will have something like

| command |

some random text

| result |

where the bit in the middle varies but the result remains more consistent. So you should be able to treat that command result pair as a sort of sublanguage.

Like as a preliminary consistency check that I just performed right, I basically ran the same prompt and then did a couple of checks that maybe show confusing behavior if it's not just smearing popular text.

I asked it for a fresh Linux installation then checked that golang wasn't installed (it wasn't). However, when I ran find / -name go, it found a Go directory (/usr/local/go) but when I run "cd /usr/local/go" also tells me I can't cd into the directory since no such file exists which would be confusing behavior if it wasn't just capturing correlations and actually understanding what find does.

I "ls ." the current directory (for some reason I was in a directory with a single "go" directory now despite never having cd'ed to /usr/local) but then ran "stat Documents/" and it didn't tell me the directory didn't exist which is also confusing if it wasn't just generating similar output to the internet.

I asked it to "curl -Z http://google.com" (-Z is not a valid option) and it told me http is not a valid protocol for libcurl. Funnily enough, running "curl http://google.com" does in fact let me fetch the webpage.

I'm a bit suspicious that the commands that the author ran are actually pretty popular so it can sort of fuzz out what the "proper" response is. I would argue that the output appears mostly to be a fuzzed version of what is popular output on the internet.

Re: GPT-4 Designed a Programming Language

#327

Earlier quoted context omitted.

That is literally what LLMs do though. Those comments try to explain how it works so you better can see how it produced the results you see, they aren't trying to say that the program is worthless just that it isn't the magic some people think it is. These models are trained to produce text snippets that look like text snippets it has seen before, and it has seen all internet. That means it can do a lot of impressive…

These comments are deceptive. Yes, this is how LLMs work, but that doesn't mean they only repeat things they have seen before. LLMs are capable of following instructions to construct new words in any language they know, words never seen before. I've seen it being dumb in maths or real world problems. But as a large language models, they understand and speak languages fine, and even mistakes they make look like mistak…

Ok, since you called out the gp comment as 'deceptive', I in turn am going to call out your comment (and others like it) as delusional, and point to specific places in your comment that exhibit this state of delusion (about LLMs).

> they understand and speak languages fine

No, they neither 'understand' nor 'speak' languages. The first word here is the more delusional, they have no understanding of languages. They have simply generated a model of the language. And they do not 'speak' the language they have modeled; they generate text in that language. Speaking generally implies an active intelligence; there is no intelligence behind an LLM. There is simply a machine generating (wholly derivative) output text from input text.

> We may as well say that when we speak, we are just predicting words we have trained on

This is the delusion, commonly being repeated, that humans themselves are only LLMs. This is a dangerous delusion, in my view, and one that has no evidence behind it. It is only a supposition, and a sterile and nihilistic one in my view.

> The general knowledge and thinking of these models are surely limited [...] I think it is very possible to break the barriers very soon

The limitations are fundamental to LLMs. LLMs have no general knowledge and LLMs don't do any 'thinking'. Your understanding of what they are doing is in grave error, and that error is based on a personification of the machine. An error coming from the delusion that because they generate 'natural' language they are operating similarly to us (false and addressed above). They are never going to break the limits because they have never started to transcend those limits in the first place, nor can they. They don't and will never 'think' or 'reason'.

Re: GPT-4 Designed a Programming Language

#328

Earlier quoted context omitted.

It also had a program giving it potential moves it could play, and explained to it when it had won the game. That isn't zero input. We don't have such programs for logic or text. So there will be no GPT zero, at least until we have already solved logic.

What is your point out of interest?

Not the person who you posed the question to, but I believe their point may be that Go is a fully described game, a constrained 'space' even if that space is remarkably deep.

Our world is not a fully described game (unless you want to take a radical philosophical position, and in which case we still don't know the rules of this game). We talk about and think about this world in language. Training a machine on human language and then after training, having that machine take language inputs and generate language outputs is not the same as training a machine to 'play' Go because the two worlds they have been trained in (1. the game of Go; 2. the real world) are fundamentally not equivalent.

Re: GPT-4 Designed a Programming Language

#329

Last year, when Blake Lemoine was fired by Google for claiming his AI was sentient, a lot of people on this website sensibly noted that the danger of these chatbots was not that they were sentient, but that average people could easily be tricked into thinking that they were. And now here we are: it seems we tech people are every bit as susceptible to this kind of thinking as the average joes we looked down on. Actual…

99.9999 percent of humans are no more than copiers of other people's ideas. With no conceptual insight, no original ideas, etc etc. That doesn't mean they're not sentient. :-)

This is (arrogant) misanthropy, and I see far too much of it on forums that contain significant numbers of people who consider themselves more 'enlightened' than the common man (reddit being a prominent example).

Re: GPT-4 Designed a Programming Language

#330

Earlier quoted context omitted.

Lol no. That's just what you tell yourself to feel smart. They're are many who understand that don't think of it as dumb. "I think GPT-3 is artificial general intelligence, AGI. I think GPT-3 is as intelligent as a human. And I think that it is probably more intelligent than a human in a restricted way… in many ways it is more purely intelligent than humans are. I think humans are approximating what GPT-3 is doing, n…

A founder trying to sell a product by telling you the product is magic? Yeah, very trustworthy source there.

That quote reveals either that they are 1) delusional or 2) a fraudster. Your comment leans more towards 2) but I think 1) is also a possibility, which in some ways I find more concerning (delusional cult leaders being potentially more dangerous than common fraudsters).
Post reply on HN