Comment by RandomLensman
3 years ago
LLMs cannot reason or use mathematics - in a way, they don't know what they are talking about. Why would such technology lead to superhuman smarts?
3 years ago
LLMs cannot reason or use mathematics - in a way, they don't know what they are talking about. Why would such technology lead to superhuman smarts?
LLMs are trained with a form of imitation learning, they imitate human (and other) text. It seems indeed not likely that pure LLMs will advance far beyond human ability, since even a perfect LLM could only imitate human text perfectly. But other approaches will follow.
What is your definition of “reasoning” here? They are clearly able to do many things that we would call reasoning if a human did them
I mention this all the time.
I wanted GPT to give me an NGINX config for Active Collab as we're previously using Apache, and thus an htaccess file. I fed it all the documentation from Active Collab, I couldn't get anything valid out of it. Hallucinated all kinds of things that weren't there. I then gave it the URL rewrites that would be required, line-by-line, spent a long time trying to correct it. No bueno, even worse hallucination. I spent days on trying to get it to output a valid NGINX config that incorporated these URL rewrites. It can't reason, it's doing exactly what LLMs do, which is next word prediction.
I can't imagine what people are using it for in terms of a valuable addition to their workflow with how much it hallucinates. If it can't even do an NGINX config, what use is it for anything else? People saying it's helping them learn programming languages. Fuck me, they don't know when it's wrong, and it will be wrong at some point, it's an LLM.
There are flaws but also consider PEBCAK error.
For one next time when it starts hallucinating and a gentle course correction doesn't do it, just start a new chat with a different prompt approach. Having the error in its context reinforces the same mistake and sometimes it can't get out of this loop.
5 replies →
It’s true that their training goal sometimes rewards them hallucinations, encouraging them do that, but that doesn’t mean it’s not reasoning. If anything hallucinations involve more reasoning because they have to make up something new that wasn’t there
12 replies →
>If it can't even do an NGINX config, what use is it for anything else?
Turn it around. I'm sure there are lots of people around who couldn't do it either, even with a ridiculously high amount of time to do it
2 replies →
The vast majority of humans would fail your reasoning test.
3 replies →
They seem to not be able to use concepts they can "explain" (beyond having answers sufficiently dense in the training set, it seems).
Humans also take a while to learn how to use concepts that they might be able to remember an answer to, though. Also keep in mind that LLMs have a limited time to produce each token and can’t “stop to think” like a human could. Obviously LLMs aren’t as good as reasoning as humans but they clearly do some (possibly quite limited) form of reasoning
7 replies →
If you put it this way, solving a differential equation numerically also provides an answer similar to human. But you don’t call Matlab an AI.
I guess that can called a form of reasoning, LLMs are much more general in their reasoning abilities though
Who knows what there will be besides LLMs. I don't really get why AI can't exceed the human brain in everything someday unless you are religious and see some supernatural aspects to the brain
Maybe we can or cannot build such a thing. We have no natural example for some exponentially self improving intelligence.
We also cannot build living animals from scratch or are anywhere close to it - maybe some forms of AI are much tougher to do than we think.
If you take a step back and look at computing in general as some amorphous evolving entity, it can be said the Machines are getting better and I would be surprised if it wasn’t exponential.
Talking out of my ass here, but my point is that I think The Machines(c) don’t look like biological and separated entities. I think it’ll look more like what we call corporations (hive minds) composed of a vast variety of different functional parts.
There was one article I read that discussed an "AI Winter". The tl;dr being that our entire global compute likely isn't sufficient enough for a hypothetical AGI.
What's more likely, IMO, is using real brain cells. https://www.ucl.ac.uk/news/2022/oct/human-brain-cells-dish-l...
However, real brain cells are a big question of ethics if it's thus actually able to think. I would argue that we've then created a slave rather than a machine, and that is unacceptable.
would like to second your observation…
AGI won’t be an LLM same as it won’t be an LSTM or CNN. But it’s an impressive step towards AGI.
Honest question: How do you know it is a step towards AGI?
If you accept that AGI is possible at all…
How can a something that generates such a massive surge of interest, investment and research into AI not be a step toward it?
Saying it’s not a step towards AGI is basically saying AGI isn’t possible at all, because it means that all our efforts are making zero progress on AGI. That’s not a falsifiable position to take.
If you’re serious, the parent post literally said “AGI isnt going to look like this”.
…but realistically, how would a LLM that could easily refine itself from experiences, and had a very large context, let’s say, a billion tokens, be meaningfully different from AGI?
It could learn. It could remember things. It could generate human like output from a complex context.
Sure, it’s just a stochastic parrot… but if it can refine the model from real world inputs (learn new tricks, learn games, etc) and generate large scale (entire books worth) of coherent conversation and interactions… where do you draw the line between that and actual AGI?
Large contexts (35k tokens) are here right now. Refining models is here right now. They’re just expensive and slow (inference and training).
Maybe the current architecture doesn’t scale up beyond that and it’s a dead end, but my gosh.
If you don’t think what we have is a step towards AGI you really have to work hard to make your definition of AGI very very difficult to attain.
4 replies →
Because a research paper claimed that they believe it's a basic and incomplete AGI. However, said paper then goes on to actually say LLMs aren't the way forward if people bother to read it.
One comment on HN called it a "baby AGI" after linking to the paper.
Eye roll inducing.
1 reply →
Current-generation LLMs give many people an impression of reasoning. I think reasoning about things is a big part of AGI.