Hate to break it to you but quality of data isn’t the fundamental problem with LLMs. It’s that they are trying to use statistics to encode entire thought processes into hidden variables from conversation snippets. They want to use statistics to go from many individual interactions to a large model, and then use that model to predict individual interactions again. Which you can do with statistics, but it’s predicting the average text that follows the prompt, not the correct text (it has no concept of correctness; whenever it “talks” about it, that’s just the average text that follows, not any particular insight into what’s correct or even how it works).
That’s not to say that the quality of the training data has no impact; it can have a huge impact. I’m just saying that even if the training data was perfect, the LLM will still get things wrong in its output.
I listened to a podcast with a couple smart mathematicians talking about AI recently and this rings true based off what I heard them discuss.
They hypothesized that only verifiable domains can really see advances due to AI. So mathematics, physics, a load of the other sciences, and medical research. Even programming, as long as you have a pre-designed solution.
But for problems where you can’t look at a solution and say “yeah, that’s an optimal solution or close to it”, ie basically any business problem; they are much less useful, a big reason being what you mentioned in your comment.
It’s that they are trying to use statistics to encode entire thought processes into hidden variables from conversation snippets. They want to use statistics to go from many individual interactions to a large model, and then use that model to predict individual interactions again.
Has it been shown that the human brain doesn’t model the world in a similar way, though? A huge portion of human knowledge is both stored and transmitted in the form of language. Lots of human knowledge also follows the garbage in, garbage out theory, where you can have entire areas of knowledge that aren’t actually true but might be internally consistent, at least within certain scopes: conspiracy theories, belief in the supernatural, entire academic disciplines built on a religion or theology that not everyone believes, etc. Or even world building in fiction, the words on a page can be enough to convey ideas such that it “tricks” human brains into filling in the gaps so that they internally see a rich, fleshed out world that is entirely fictional and where specific details might not find strong direct support in the underlying text.
it has no concept of correctness
But statistical weight on what is more or less likely to be correct still makes a difference to objective quality of the outputs. If the model weights are trained on the reality that high quality university texts describe something and reflect some sort of underlying model of what is described using language, then can’t the model itself learn as much as a human could from those words on a page?
All models are wrong, but some can be useful. And different models have different quality in different domains. So although I don’t believe LLMs will overtake the hump of getting ahead of human knowledge, I also don’t believe that any given LLM can be evaluated on quality, and that Facebook’s LLMs are significantly behind other LLMs we see.
And that maybe a huge part of it is its internal process of preparing the model to evaluate the quality of its inputs, such that the output it produces can also score high on quality.
But it isn’t encoding knowledge, it’s encoding word correlations. That’s how it can get things wrong like saying fat32 won’t be good for a 64GB removable drive because fat32 only has a 2TB address space.
Or how it can get something wrong and when you point it out, it immediately sees how it was wrong. And I realize that that sounds human, but the way it gets there is very different. It’s predicting responses based off word correlations, not using knowledge recall to apply facts and relations known about the topics and generate responses from that.
Hate to break it to you but quality of data isn’t the fundamental problem with LLMs. It’s that they are trying to use statistics to encode entire thought processes into hidden variables from conversation snippets. They want to use statistics to go from many individual interactions to a large model, and then use that model to predict individual interactions again. Which you can do with statistics, but it’s predicting the average text that follows the prompt, not the correct text (it has no concept of correctness; whenever it “talks” about it, that’s just the average text that follows, not any particular insight into what’s correct or even how it works).
That’s not to say that the quality of the training data has no impact; it can have a huge impact. I’m just saying that even if the training data was perfect, the LLM will still get things wrong in its output.
I listened to a podcast with a couple smart mathematicians talking about AI recently and this rings true based off what I heard them discuss.
They hypothesized that only verifiable domains can really see advances due to AI. So mathematics, physics, a load of the other sciences, and medical research. Even programming, as long as you have a pre-designed solution.
But for problems where you can’t look at a solution and say “yeah, that’s an optimal solution or close to it”, ie basically any business problem; they are much less useful, a big reason being what you mentioned in your comment.
Has it been shown that the human brain doesn’t model the world in a similar way, though? A huge portion of human knowledge is both stored and transmitted in the form of language. Lots of human knowledge also follows the garbage in, garbage out theory, where you can have entire areas of knowledge that aren’t actually true but might be internally consistent, at least within certain scopes: conspiracy theories, belief in the supernatural, entire academic disciplines built on a religion or theology that not everyone believes, etc. Or even world building in fiction, the words on a page can be enough to convey ideas such that it “tricks” human brains into filling in the gaps so that they internally see a rich, fleshed out world that is entirely fictional and where specific details might not find strong direct support in the underlying text.
But statistical weight on what is more or less likely to be correct still makes a difference to objective quality of the outputs. If the model weights are trained on the reality that high quality university texts describe something and reflect some sort of underlying model of what is described using language, then can’t the model itself learn as much as a human could from those words on a page?
All models are wrong, but some can be useful. And different models have different quality in different domains. So although I don’t believe LLMs will overtake the hump of getting ahead of human knowledge, I also don’t believe that any given LLM can be evaluated on quality, and that Facebook’s LLMs are significantly behind other LLMs we see.
And that maybe a huge part of it is its internal process of preparing the model to evaluate the quality of its inputs, such that the output it produces can also score high on quality.
But it isn’t encoding knowledge, it’s encoding word correlations. That’s how it can get things wrong like saying fat32 won’t be good for a 64GB removable drive because fat32 only has a 2TB address space.
Or how it can get something wrong and when you point it out, it immediately sees how it was wrong. And I realize that that sounds human, but the way it gets there is very different. It’s predicting responses based off word correlations, not using knowledge recall to apply facts and relations known about the topics and generate responses from that.