>> The raw version of GPT apparently achieves an Elo score of 1800 at chess: https://twitter.com/GrantSlatton/status/1703913578036904431 https://twitter.com/Gra

>> The raw version of GPT apparently achieves an Elo score of 1800 at chess: https://twitter.com/GrantSlatton/status/1703913578036904431 https://twitter.com/GrantSlatton/status/1703913578036904431. This is worse than a good dedicated chess engine, but almost certainly better than I could do! Probably because it was trained to get better at it. https://twitter.com/GrantSlatton/status/1706545268009931175 https://twitter.com/GrantSlatton/status/1706545268009931175 There's already been plenty of discussion on LLMs' ability to play chess and it makes sense that OpenAI would try to make their models better at it.