Comment by redox99
12 hours ago
Most programming tasks are exactly like that. Is this agent able to complete this task? Is this agent able to optimize a kernel beyond previous attempts?
Of course some are subjective and that's where progress is harder, like "Is this website pretty?". But for tasks that can be objectively measured, LLMs will go beyond human level, just like with Chess and Go.
That's why RL is so important when training LLMs.
My point is that LLMs depend on training data so the code they produce will be stuck in 2022, no new languages, techniques beyond that because new techniques are not in the training data (at least not enough of it for training because most coders are now using LLMs).
Chess/Go continues to progress because it is primarily a human vs human activity, people will always be learning to play chess and chess will continue to develop.
> Chess/Go continues to progress because it is primarily a human vs human activity, people will always be learning to play chess and chess will continue to develop.
AIs are not continuing to get better at chess/go because humans continue to play at levels far below themselves who discover new techniques. They get better because they play against other AIs and discover new techniques that have a higher win rate that way.
I would bet that even if humans stopped playing chess/go and people were still willing to run these AI models against each other they would continue to get better.
I am not talking about the advancement of AI, I am talking about the advancement of chess.
Two things can be true AI drastically contribute to the advancement of chess and humans playing against each other also contribute (even if slowly) to the advancement of chess as it has always been since the invention of the game. The point is that because chess is primarily a human vs human game humans will always have the knowledge of chess, unlike with programmers who are giving it up to prompting, and programming being much more complex than chess (checkmate and win) will be stuck in 2022 because of the training data.
Pre training data is in large part synthetic these days, and RL data is almost all synthetic.
Computer Chess progress has nothing to do with human vs human activity. AlphaGo Zero used no human game data at all.
> Pre training data is in large part synthetic these days
How much of that data can lead to innovation? Can you predict all innovation map it out on paper.
> Computer Chess progress has nothing to do with human vs human activity.
The point is that humans will always be learning chess because it primarily a human vs human activity they will be contributing games to the chess database, unlike with programmers who are stopping to code and only prompting, generating code stuck in 2022.
> AlphaGo Zero used no human game data at all.
Sure, but that instance of AlphaGo is still dependent on its training, its intelligence, so it is a question of is that the best and only way to win a game of Go. Just a few weeks ago, a Go Grandmaster found a way to beat one of the strongest Go AIs.
So a specific instance of an LLM might be the smartest based on what we know and need today but that is not the limit of how far we can go, this is why it is important for humans to always have an intimate connection with the code, math, science, chess etc for progress to continue.
8 replies →