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

Life is more than an engineering problem

lareviewofbooks.org

Post reply on HN
Print view View original

111–120 of 257 posts

Re: Life is more than an engineering problem

#111

Post by rileymat2 »

sleepytree wrote: post #85
abecedarius wrote: post #68

Chiang makes some insightful points, e.g. about what we mean by magic. Then I come to > [LLMs] can get better at reproducing patterns found online, but they don’t become capable of actual reasoning; it seems that the problem is fundamental to their architecture. and wonder how an intelligent person can still think this, can be so absolute about it. What is "actual" reasoning here? If an AI proves a theorem is it only…

If I write a book that contains Einstein's theory of relativity by virtue of me copying it, did I create the theory? Did my copying of it indicate anything about my understanding of it? Would you be justified to think the next book I write would have anything of original value? I think what he is trying to say is that LLMs current architecture seems to mainly work by understanding patterns in the existing body of kno…

Why can’t creativity be taking the works, a bunch of works, finding a pattern then randomly perturbing a data point/concept to see if there are new patterns.

Then cross referencing that new random point/idea to see if it remains internally consistent with the known true patterns in your dataset.

This is how humans create new ideas often?

Re: Life is more than an engineering problem

#112

Post by munksbeer »

southernplaces7 wrote: post #110
abecedarius wrote: post #68

Chiang makes some insightful points, e.g. about what we mean by magic. Then I come to > [LLMs] can get better at reproducing patterns found online, but they don’t become capable of actual reasoning; it seems that the problem is fundamental to their architecture. and wonder how an intelligent person can still think this, can be so absolute about it. What is "actual" reasoning here? If an AI proves a theorem is it only…

I see absolutely zero wrong with that statement. What he said is indeed much more reasoned and intelligent than the average foolish AI hype i've often found here, written by people who try to absurdly redefine the obvious, complex mystery that is consciousness into some reductionist notion of it being anything that presents the appearance of reasoning through technical tricks. Chiang has it exactly right with his dou…

> Chiang has it exactly right with his doubts, and the notion that pattern recognition is little different from the deeply complex navigation of reality we living things do is the badly misguided notion.

How do you know this?

Re: Life is more than an engineering problem

#113

Post by BeetleB »

emtel wrote: post #70

I like Chiang’s stories, but he simply doesn’t know what he’s talking about when it comes to LLMs. Forget the philosophical points - he claims that LLMs are strictly worse than search engines for information retrieval. This is just obviously false, I can give you dozens of conversations from my ChatGPT history over the past two months which would have been impossible to get answers to on Google.

The comparison with search is faulty to begin with. Yes, you can search with an LLM, but that's a side effect of the tool.

While I certainly also have found things via LLMs that I couldn't easily with a search engine, the number of false positives is huge. My heuristic is:

If I ask an LLM something and it's easy to verify via Google because its answer narrows the search space - then I'll use it. Otherwise, Google is still king.

Example: Asking an LLM the health benefits of supplement X is a waste of time. Verifying everything it tells me would be the same amount of work as asking a search engine.

Example: Asking how to solve a given coding problem is great, because it drastically reduces the search space. I only have to look up the particular function/API calls it uses.

Ditto for asking how to achieve a task in the command line - I can quickly verify the arguments are accurate via the man page.

Most of the things I search for do not fall into this category, but in the category of "still need to do the same amount of work as just searching via Google."

Re: Life is more than an engineering problem

#114

Post by BeetleB »

teleforce wrote: post #13

> And even though I know a perfect language is impossible It's already existed for a very long time and it's called Arabic language. It's the extreme opposite of English where English is a hodgepodge of a languages mixtures where about 1/3 is French language, about one third is old English and about 1/3 of other world's languages including Arabic. Comparing the best of English literatures for example Shakespeare's bo…

Really? OK, here goes with my litmus test of languages:

How do you say the plural you when referring to a group of males?

How do you say the plural you when referring to a group of females?

How do you say the plural you when referring to a group of both males and females?

If you don't have a different phrasing for each, then it is imperfect.

Re: Life is more than an engineering problem

#115

Post by jrflowers »

abecedarius wrote: post #68

Chiang makes some insightful points, e.g. about what we mean by magic. Then I come to > [LLMs] can get better at reproducing patterns found online, but they don’t become capable of actual reasoning; it seems that the problem is fundamental to their architecture. and wonder how an intelligent person can still think this, can be so absolute about it. What is "actual" reasoning here? If an AI proves a theorem is it only…

This is a good point.

If you prick an LLM does it not bleed? If you tickle it does it not laugh? If you poison one does it not die? If you wrong an LLM shall it not revenge?

Re: Life is more than an engineering problem

#116

Post by southernplaces7 »

vouwfietsman wrote: post #97
sleepytree wrote: post #85

Earlier quoted context omitted.

If I write a book that contains Einstein's theory of relativity by virtue of me copying it, did I create the theory? Did my copying of it indicate anything about my understanding of it? Would you be justified to think the next book I write would have anything of original value? I think what he is trying to say is that LLMs current architecture seems to mainly work by understanding patterns in the existing body of kno…

> But it is clear humans are capable of ... Its not though, nobody really knows what most of the words in that sentence mean in the technical or algorithmical sense, and hence you can't really say whether llms do or don't possess these skills.

>nobody really knows what most of the words in that sentence mean in the technical or algorithmical sense

And nobody really knows what consciousness is, but we all experience it in a distinct, internal way that lets us navigate the world and express ourselves to others, yet apparently some comments seem to dismiss this elephant of sensation in the room by pretending it's no different than some cut and dried computational system that's programmed to answer certain things in certain ways and thus "is probably no different from a person trained to speak". We're obviously, evidentially more than that.

Re: Life is more than an engineering problem

#117

Post by abecedarius »

fragmede wrote: post #84
ustad wrote: post #77

Earlier quoted context omitted.

Has an AI “proven” a theorem?

Leading mathematician Terence Tao says yes, with a lot of guidance. https://mathstodon.xyz/@tao/113132503432772494

I have a similar opinion of Claude Sonnet. Superhuman knowledge; ability to apply it to solve new math/coding problems at roughly the level of a motivated high-schooler (but not corresponding exactly in profile to anything human); less ability to stay on track the longer the effort takes.

But ChatGPT a couple years ago was at more like grade-school level at problem-solving. What should I call this thing that the best LLMs can do better than the older ones, if it's not actual reasoning? Sparkling syllogistics?

Sorry, that's sarcastic, but... it's from a real exasperation at what seems like a rearguard fight against an inconvenient conclusion. I don't like it either! I think the rate of progress at building machines we don't understand is dangerous. (Understanding the training is not understanding the machinery that comes out.)

Compare the first previews of Copilot with current frontier "reasoning" models, and ask how this will develop in the next five years. Maybe it'll fizzle. If you're very confident it will: I'd like to be convinced too.

Re: Life is more than an engineering problem

#118

Post by photochemsyn »

> “It’s like imagining that a printer could actually feel pain because it can print bumper stickers with the words ‘Baby don’t hurt me’ on them."

Problem: the human brain has no pain receptors, no nociceptors. It just takes in messages from remote nerves and 'prints a bumper sticker' that tells higher cognitive centers 'you're feeling pain!'. What's the difference?

> "LLMs are like a search engine that rephrases information instead of giving it verbatim or pointing you to the original source."

Problem: How does this differ from human learning? If a human reads a book and tells someone else about it, constructs a summary of the important points and memorable passages, how is that fundamentally different from what LLMs are doing?

The second one really impacts the intellectual property arguments - if training a model on data is fundamentally similar to training a human on data, does 'derivate work' really apply to the creations of the human or of the model?

Re: Life is more than an engineering problem

#119

Post by justinpombrio »

m_fayer wrote: post #36
vrnvu wrote: post #4

Highly recommend "Stories of Your Life and Others". I describe Ted Chiang as a very human sci-fi author, where humanity comes before technology in his stories. His work is incredibly versatile, and while I expected sci-fi, I'd actually place him closer to fantasy. Perfect for anyone who enjoys short stories with a scientific, social, or philosophical twist. Another anthology I'd recommend with fresh ideas is Axiomati…

While he’s very much unique, the one writer he brings to my mind is Borges, just a bit more humane and steeped in real technology and theory.

Chiang writes science-fiction, Egan writes physics-fiction, and Borges wrote philosophy-fiction.

Re: Life is more than an engineering problem

#120

Post by fragmede »

abecedarius wrote: post #117
fragmede wrote: post #84

Earlier quoted context omitted.

Leading mathematician Terence Tao says yes, with a lot of guidance. https://mathstodon.xyz/@tao/113132503432772494

I have a similar opinion of Claude Sonnet. Superhuman knowledge; ability to apply it to solve new math/coding problems at roughly the level of a motivated high-schooler (but not corresponding exactly in profile to anything human); less ability to stay on track the longer the effort takes. But ChatGPT a couple years ago was at more like grade-school level at problem-solving. What should I call this thing that the best…

you said you said it sarcastically but I like "syllogistic" a lot. We need more volcabulary to describe what LLMs do, and if I tell ChatGPT A implies B implies C, and I tell it A is true, and I can describe that as the LLM syllogisting and not use the words "reasoning" or "thinking", that works for me.

As far as if it will fizzle, even if it does, what we have currently is already useful. Society will take time to adjust to ChatGPT-4's level of capabilities, nevermind whatever OpenAI et al releases next. It can't yet replace a software engineer, but it makes projects possible they previously weren't attempted because they required too much investment previously. So unless you're financially exposed to AI directly (which you might be, many people are!), the question of if it's going to fizzle is more academic than something that demands a rigorous answer. Proofs of a negative are really hard. Reusable rockets were "proven" to be impossible right up until it was empirically proven possible.

Post reply on HN