Comment by merelydev
15 hours ago
> 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.
> 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.
If this were true then it would be impossible for these models to ever exceed the top human level as there would exist no training data that allows them to exceed the top human level.
However, despite there being no training data on ability to beat the top humans these models have achieved it.
> 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.
This is just you wanting to remain relevant rather than actually based on evidence.
> If this were true then it would be impossible for these models to ever exceed the top human level as there would exist no training data that allows them to exceed the top human level.
Of course AI exceeds humans at chess, I never denied that. I am saying because chess is primarily a human vs human game, humans will always be learning and playing chess, their games will add to the chess knowledge base, AI also adds to this knowledge base. But programming is not primarily a human vs human activity so there is a risk programmers will forget how to code and all software will be stuck in 2022 because of the training data, this stifles innovation.
> I am saying because chess is primarily a human vs human game, humans will always be learning and playing chess, their games will add to the chess knowledge base
I guess I'm contesting that idea you are putting forward that the data from the games these humans are playing, which are at a vastly lower level that the top AIs are meaningfully important for helping the AIs to improve at the top level.
Would more people learning their times tables be helpful for top mathematicians in their fields to get better at the frontier of maths? Probably not right. Same applies here.
Would AIs advance at the same rate for the top level of chess in a world where humans completely stopped playing chess vs the world we have today. I would say they would as the human level data is of limited value to the frontier which is dominated by AI and AI game data, you are claiming that it does.
> But programming is not primarily a human vs human activity so there is a risk programmers will forget how to code and all software will be stuck in 2022 because of the training data, this stifles innovation.
Does it? Or will AI be able to run its own experiments and find better/more efficient abstractions that propagate because they are better and this will find its way into training data for future AI.
3 replies →
Human games databases are completely irrelevant to the strongest chess engines. We are ants in comparison. The Go thing you mention is playing against handicap. Sorry I won't go into more detail explaining why your premises are wrong, I'm tired of this discussion.
Human games databases might not contribute to AI training but they do contribute to the chess database itself like it has always been, AI doesn't change that fact.