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

AI isn’t good enough

skventures.substack.com

Post reply on HN
Print view View original

161–170 of 374 posts

Re: AI isn’t good enough

#161

Post by pdimitar »

garblegarble wrote: post #158
nine_k wrote: post #137

Earlier quoted context omitted.

The current crop of LLMs is like Paleozoic megafauna, or like Egyptian Pyramids. It takes a few relatively simple approaches and stretches them wildly, using colossal computing resources. Live systems in nature seem to solve similar problems with way less compute available. There should be better architectures. Also, as somebody said, every exponential growth curve is a lower part of a sigmoid. LLMs will plateau at s…

>Live systems in nature seem to solve similar problems with way less compute available Do they really? They're certainly more energy-efficient in business-as-usual mode, but a human brain has 86 billion neurons, 600+ trillion synapses(!), and each instance takes 15-20+ years to train to do complex logical tasks. Even if the per-cell work is tiny (and, is it? cells are amazingly complex), 86 billion (or 600+ trillion)…

You are ignoring the fact that a toddler, once their musculature develops, is able to learn to walk after several tries.

Show me a humanoid robot that can do that.

Re: AI isn’t good enough

#162

Post by the8472 »

RandomLensman wrote: post #155
the8472 wrote: post #152

Earlier quoted context omitted.

It's like if a squirrel started playing chess and instead of "holy shit this squirrel can play chess!" most people responded with "But his elo rating sucks"

I don't understand why anyone was surprised by computers processing and generating language or images.

There are many reasons. Failing at extrapolating exponentials. Uncertain thresholds for how much compute and data each individual task requires. Moravec's paradox, and relatedly people expecting formalizable/scientific problems to be solved first before arts. There are still some non-materialists. And a fairly basic reason: Not following the developments in the field.

Re: AI isn’t good enough

#163

Post by dontupvoteme »

HPsquared wrote: post #116
dontupvoteme wrote: post #61

Earlier quoted context omitted.

That's where good money and societal upheaval will come up, private models for corporations trained on their entire codebase and documentation. Both young and old shall feel the pain then.

The codebase and documentation usually don't capture the full context of the development process. There is still a lot that isn't written down.

I agree, but the MBAs will probably not.

Re: AI isn’t good enough

#164

Post by stared »

usrbinbash wrote: post #67
ambrozk wrote: post #18

This entire piece is based on one massive, unsupported assertion, which is that LLM progress will cease. Or, as the author puts it, "we are at the tail end of the first wave of large language model-based AI... [it] ends somewhere in the next year or two with the kinds of limits people are running up against." I want to know only one thing, which is what gives him the confidence necessary to say that. If that one stat…

> This entire piece is based on one massive, unsupported assertion, which is that LLM progress will cease. Which is countered by...the assertion that it won't? LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. They can be fine tuned to specific tasks, but at their core, they remain stochastic parrots. > I want to know only one thing, which is what gives him the confide…

> LLMs won't get intelligent. That's a fact based on their MO. They are sequence completion engines. They can be fine tuned to specific tasks, but at their core, they remain stochastic parrots.

Yet, we are different, right?

Re: AI isn’t good enough

#165

Post by dontupvoteme »

jiggawatts wrote: post #99
dontupvoteme wrote: post #49

Earlier quoted context omitted.

Why not "simply" multigen every (important) query and take the statistical average? Hallucinations are random, the truth isn't. This is absurdly expensive with GPT4, cheaper with 3, and dirt cheap locally with LLaMA

That’s an established technique with papers written on the topic and everything. Anecdotally I tested this by having GPT4 translate Acadian cuneiform — which it can just barely do. I had it do this four times and it returned four gibberish answers. I then prompted it with the source plus the four attempts and asked for a merged result. It did it better than the human archeologists did! More readable and consistent. I…

Makes sense, it's too obvious to not have already been studied :) Do you know what I might search for to find info on it? >Expensive now… soon to be standard? With translation, better safe than sorry? It's a very important field and preserves human history so, why not?

Re: AI isn’t good enough

#166

Post by Drakim »

pdimitar wrote: post #161
garblegarble wrote: post #158

Earlier quoted context omitted.

>Live systems in nature seem to solve similar problems with way less compute available Do they really? They're certainly more energy-efficient in business-as-usual mode, but a human brain has 86 billion neurons, 600+ trillion synapses(!), and each instance takes 15-20+ years to train to do complex logical tasks. Even if the per-cell work is tiny (and, is it? cells are amazingly complex), 86 billion (or 600+ trillion)…

You are ignoring the fact that a toddler, once their musculature develops, is able to learn to walk after several tries. Show me a humanoid robot that can do that.

To be fair a toddler's brain isn't "offline" while it's musculature develops, they kick and move their limbs constantly, and then later crawl and and even stand, before they finally take the first walking steps. It's just objectively wrong to say a toddler goes from zero to walking in "several tries".

Re: AI isn’t good enough

#167

Post by M4v3R »

pdimitar wrote: post #161
garblegarble wrote: post #158

Earlier quoted context omitted.

>Live systems in nature seem to solve similar problems with way less compute available Do they really? They're certainly more energy-efficient in business-as-usual mode, but a human brain has 86 billion neurons, 600+ trillion synapses(!), and each instance takes 15-20+ years to train to do complex logical tasks. Even if the per-cell work is tiny (and, is it? cells are amazingly complex), 86 billion (or 600+ trillion)…

You are ignoring the fact that a toddler, once their musculature develops, is able to learn to walk after several tries. Show me a humanoid robot that can do that.

> You are ignoring the fact that a toddler, once their musculature develops, is able to learn to walk after several tries.

It's definitely not several tries. I've actually witnessed that very recently with my daughter. It takes a child literally several thousand attempts and falls (around a hundred falls per day) and 1,000 hours of dedicated practice, before they can make their first step without holding anything [1]. And a lot more until they can walk reliably.

[1] Source: Becoming You, S1E2, from 21:30

Re: AI isn’t good enough

#168

Post by ozgung »

dontupvoteme wrote: post #45
system2 wrote: post #36

Earlier quoted context omitted.

For what it's worth, I can actually tell the hype around AI died down a little. 8-10 months ago you'd hear AI from grandmas and their interest in playing with gpt. That's gone. From programmer's perspective, it is still useless. I can't still get chatgpt to give me a useful ffmpeg or a basic php database script. I don't know who is really using ai for what, other than crappy chatbots. They are saying JSON works perfe…

Really? I have a lot of success with bash and ffmpeg scripts. I don't even try to write them myself these days. I don't do JSON stuff much but either function calling or grammar based sampling should be able to force correct formatting if the issue is syntax. ImageGen seemed like a solved problem to me with stable diffusion about 6 months ago or so, between LoRAs and the tons of models availble on sites like civitas.…

Same. GPT-4 completely replaced my googling/stackoverflow routine. I generates a customized solution for my specific needs with step by step instructions. I think it works in a smarter (but slower) mode if the question is non-trivial. With pdf plugins it helps me reading and understanding recent papers. It generates an excellent summary and answers all my questions regarding any advanced concepts in the paper. I believe it's already smarter than me. I don't know if it can replace me in the mid-term but it definitely increases my productivity.

Re: AI isn’t good enough

#169

Post by RandomLensman »

w1nk wrote: post #126
nopinsight wrote: post #108

Earlier quoted context omitted.

Geoffrey Hinton, Andrew Ng, and quite a few other top AI researchers believe that current LLMs (and incoming waves of multimodal LFMs) learn world models; they are not simply 'stochastic parrots'. If one feeds GPT-4 a novel problem that does not require multi-step reasoning or very high precision to solve, it can often solve it.

Just to further this, it's not just 'big names' that feel this way. Read this paper from a team at Microsoft Research: https://arxiv.org/abs/2303.12712 . These folks spent months studying properties of GPT-4, that paper is ~150 pages of examples probing the boundaries of the model's world understanding. There is obviously some emergent complexity arising from the training procedure.

That paper makes some pretty strong claims in the abstract that are not all really supported by the body of the paper. For example, there isn't much on the law or medicine claims in the paper.

Re: AI isn’t good enough

#170

Post by Ekaros »

How far are we from a point you could leave AI to prompt something itself. That is not just respond to something, but actively seek it.

Let's say use this place. Could it without being ordered read articles, comments and add it own comments? Maybe search internet by itself for relevant content and post it? Maybe try to earn maximal amount of upvotes? And all this without there being any bit of code poking it to do it?

Post reply on HN