7 ms·

> I haven’t fully verified the proof; I worked through it with Claude Fable and it passes the sniff test, but fully digesting it will take a bit more energy tha

by chr15m 9d ago

> I haven’t fully verified the proof; I worked through it with Claude Fable and it passes the sniff test, but fully digesting it will take a bit more energy than I have right now. This is overlooked but really important. Nobody wants to watch robots play chess against each other - we watch human chess even though bots are better. Nobody needs math that has not been verified by humans. Human understanding and communication is the important part of mathematics.

stabbles 9d ago
> Nobody wants to watch robots play chess against each other - we watch human chess even though bots are better Many humans have studied robots play chess and learned from them. For example the concept of a "thorn pawn" was popularized after Leela played it with great success. > Nobody needs math that has not been verified by humans. There already are proofs that are understood by a handful of humans; without understanding you can still proof corollaries. For example, there are proofs to theorems that are conditional on the Riemann Hypothesis.
- chr15m 9d ago
  > Many humans have studied robots play chess and learned from them. Yes, robot chess is very useful for chess players. It's a great technology. People don't want to watch robot chess. They want to watch humans, even those who used AI to train. The human part of chess is what matters. > There already are proofs that are understood by a handful of humans Yes, exactly.
  - Mr_Minderbinder 9d ago
    > People don't want to watch robot chess. I do and I have for a number of years. Most people prefer to watch human chess but there are also those who watch computer chess with the same level of enthusiasm: https://tcec-chess.com/ https://tcec-chess.com/ https://en.wikipedia.org/wiki/TCEC https://en.wikipedia.org/wiki/TCEC
whoarewethen 9d ago
[dead]
- chr15m 9d ago
  For sure LLMs will be used more and more.

by chr15m 9d ago

stabbles 9d ago

> Nobody wants to watch robots play chess against each other - we watch human chess even though bots are better Many humans have studied robots play chess and learned from them. For example the concept of a "thorn pawn" was popularized after Leela played it with great success. > Nobody needs math that has not been verified by humans. There already are proofs that are understood by a handful of humans; without understanding you can still proof corollaries. For example, there are proofs to theorems that are conditional on the Riemann Hypothesis.

chr15m 9d ago
> Many humans have studied robots play chess and learned from them. Yes, robot chess is very useful for chess players. It's a great technology. People don't want to watch robot chess. They want to watch humans, even those who used AI to train. The human part of chess is what matters. > There already are proofs that are understood by a handful of humans Yes, exactly.
- Mr_Minderbinder 9d ago
  > People don't want to watch robot chess. I do and I have for a number of years. Most people prefer to watch human chess but there are also those who watch computer chess with the same level of enthusiasm: https://tcec-chess.com/ https://tcec-chess.com/ https://en.wikipedia.org/wiki/TCEC https://en.wikipedia.org/wiki/TCEC

whoarewethen 9d ago

[dead]

chr15m 9d ago
For sure LLMs will be used more and more.