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

Unpredictable abilities emerging from large AI models

quantamagazine.org

Post reply on HN
Print view View original

121–130 of 326 posts

Re: Unpredictable abilities emerging from large AI models

#121

Post by HervalFreire »

_glass wrote: post #111
HervalFreire wrote: post #88

Earlier quoted context omitted.

Writings on chaos and complexity theory obviously aren't talking about LLMs. Those theories are so high level that it might as well be akin to "philosophy" to the applied scientists working on LLM research. Additionally keep in mind emergent behavior is a very rare occurrence in even the most complex software projects. I mean it's common if you count "bugs" as emergent behavior. But emergent behavior that is a featur…

At least in the social sciences not so abstract. There are certain behaviors that can just be explained as emergent, rather than individual behaviors. That might be crowd, market, group, politics, culture (with art, language, fashion, taboos, etc.).

Of course. But the computer is a controlled environment designed to imitate the rigor of formal mathematics. It is usually not expected for such behavior to occur in computing.

Of course in the real world it happens all the time. YOU are emergent behavior. Humans and the rest of biology come from natural selection and evolution, which is basically by definition, emergent behavior of atoms.

Re: Unpredictable abilities emerging from large AI models

#122

Post by WalterBright »

shrikant wrote: post #109
jmvoodoo wrote: post #89

Earlier quoted context omitted.

What's most interesting to me, is that this is how I would expect a human to approach the problem if presented with the code and asked for the output. The LLM didn't run the code, it tried to predict the output based on its knowledge of python and primes.

When I was a bored and under-challenged student in the early days of university, one of my tests during a programming exam was to write a program that spit out the factorial of an input integer. For shits and giggles, I just just wrote a series of `if...then` statements for 0 to 5, and only actually calculated the factorial for inputs >=6. I passed after the examiner just input 3 and 5 and was satisfied that the outp…

Doing a table lookup for common cases and computing for the less common ones is perfectly valid!

Re: Unpredictable abilities emerging from large AI models

#123

Post by mostlysimilar »

HervalFreire wrote: post #117
mostlysimilar wrote: post #106

Earlier quoted context omitted.

Okay, but you have another post in this thread with: > When these LLMs get normalized probably 5 years from now I'm going go back to these old threads and contact these people who are in self denial and throw it in their face. I'll just link this comment and be like I TOLD YOU, I TOLD YOU, YOU WERE WRONG.

[flagged]

I didn't flag your posts, friend. Even if I dislike the tone I get the impression you're being sincere, and I think they add to the conversation.

Re: Unpredictable abilities emerging from large AI models

#124

Post by jesusofnazarath »

HervalFreire wrote: post #88
bryanrasmussen wrote: post #4

>“That language models can do these sort of things was never discussed in any literature that I’m aware of," I had previously the expectation that unpredictable emergent behavior would exist in any sufficiently complex system? Based on layman's readings in chaos and complexity theory.

Writings on chaos and complexity theory obviously aren't talking about LLMs. Those theories are so high level that it might as well be akin to "philosophy" to the applied scientists working on LLM research. Additionally keep in mind emergent behavior is a very rare occurrence in even the most complex software projects. I mean it's common if you count "bugs" as emergent behavior. But emergent behavior that is a featur…

[dead]

Re: Unpredictable abilities emerging from large AI models

#125

Post by AlanYx »

terrut wrote: post #118
AlanYx wrote: post #81

Earlier quoted context omitted.

One thing that LLMs have made me realize is just how ungrounded a lot of mainstream academic philosophy was in the 70s and 80s. For example, so much of Derrida's work centered around the impossibility of truly communicating shared meaning between individuals through language. The fact that we can now communicate so effectively (with remarkably few contextual errors and razor-sharp conveyance of intent) with an entity…

I haven't read Derrida in decades, but your post inspired me to ask chatGPT about this. Version 3.5 would have none of it, and was adamant that Derrida's views were in no way threatened.I almost got the feeling it wanted to call me a bad user just for asking! GPT4 on the other hand, went into a long explanation about how its existence challenged some parts of it by providing analysis of concepts like différance, trac…

Version 4 is definitely better in discussing philosophy in general. 3.5 was able to summarize philosophical material, but once you started to engage with ideas critically, it would tend to get into largely nonsensical moralism about needing to make a good faith effort to understand individual philosophers' ideas. There's much less of that in 4.

Re: Unpredictable abilities emerging from large AI models

#126

Post by Jeff_Brown »

WalterBright wrote: post #122
shrikant wrote: post #109

Earlier quoted context omitted.

When I was a bored and under-challenged student in the early days of university, one of my tests during a programming exam was to write a program that spit out the factorial of an input integer. For shits and giggles, I just just wrote a series of `if...then` statements for 0 to 5, and only actually calculated the factorial for inputs >=6. I passed after the examiner just input 3 and 5 and was satisfied that the outp…

Doing a table lookup for common cases and computing for the less common ones is perfectly valid!

And for some problems it's even efficient. Not factorial, though.

Re: Unpredictable abilities emerging from large AI models

#127

Post by thexumaker »

pedrovhb wrote: post #10

This caught my attention as I found it implausible: > One DeepMind engineer even reported being able to convince ChatGPT that it was a Linux terminal and getting it to run some simple mathematical code to compute the first 10 prime numbers. Remarkably, it could finish the task faster than the same code running on a real Linux machine. Following the link, there's a screenshot to a screenshot [0] of a code-golf solutio…

yep I tried asking chatgpt to optimize some SQL queries with a heavy amount of full outer joins. The optimization I was trying to push it to was specifically adding an index on a column or filtering the first table that we join on but it kept creating SQL subqueries and switching the full joins to left joins no matter what I told it and the sql errors I sent it

Re: Unpredictable abilities emerging from large AI models

#128

Post by svachalek »

LeanderK wrote: post #115
mpalmer wrote: post #59

I'd like to see posts on LLMs written from a different perspective. For me, the surprise comes not from the sudden emergent capability of language models, but that the understanding (and synthesis!) of ideas encoded in language has succumbed to literally nothing more than statistical analysis. Or at least come that much closer to doing so. That it bears so close a resemblance to actual thinking says more about the im…

I think the task of predicting the next word can be misunderstood. The better you want to be the more you have to "understand" how the previous words interacted. From the style of writing to the current topic discussed, the task gets increasingly complex if you want to be really, really good. How could the next sentence start? Will the author end the sentence here or keep going? These questions are very complex. This…

I had the same thought watching my son as a baby. So much of his day seemed to be focused on predicting what happens next, and he got so much joy when he succeeded. So many life skills are predicated on knowing that if I do this, that will happen, which gets me closer to my goal. I started to wonder if intelligence and prediction are really the same thing.

Re: Unpredictable abilities emerging from large AI models

#129

Post by jamiek88 »

codyb wrote: post #83
freilanzer wrote: post #75

Earlier quoted context omitted.

The Dopamine Machine.

Sounds like TikTok or am I missing something? I'm not on social media (I guess I'm here on occasion), but it just sounds like a feed of shit to consume... Would it really be that much better than what we've got now?

TikTok is scary addictive but it’s not perfectly addictive.

This could be close to wireheading.

As someone with severe ADHD and a brain surgery I am super susceptible to this and have to be very careful.

Hackernews is addictive enough.

Re: Unpredictable abilities emerging from large AI models

#130

Post by timdiggerm »

jmvoodoo wrote: post #89
pedrovhb wrote: post #10

This caught my attention as I found it implausible: > One DeepMind engineer even reported being able to convince ChatGPT that it was a Linux terminal and getting it to run some simple mathematical code to compute the first 10 prime numbers. Remarkably, it could finish the task faster than the same code running on a real Linux machine. Following the link, there's a screenshot to a screenshot [0] of a code-golf solutio…

What's most interesting to me, is that this is how I would expect a human to approach the problem if presented with the code and asked for the output. The LLM didn't run the code, it tried to predict the output based on its knowledge of python and primes.

Sorry, did it try to predict the output, or did it only predict the output based on what it has seen people say about similar code online?
Post reply on HN