Back in 2023 I'm pretty sure I saw people claiming that GPT-4 could play competent chess. As far as I can tell, none of today's models can go more than 5-10 moves without trying to do something illegal. Were the earlier claims wrong or have the models gotten a lot worse?
Post
Conversation
![David Watson 🥑](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
Post your reply
There's a follow up! Fun to see how all the ideas for how to improve LLM chess ability are so hit or miss
