News for hackers. Stuff that matters.

A comment on When AI Thinks It Will Lose, It Sometimes Cheats, Study Finds. Parent comment

Re:When AI Thinks It Will Lose, It Sometimes Cheats, Study F... by animal-husband on Saturday February 22, 2025 @11:07AM (#43140171)
Text generated prior to a decision to “explain” it is reasoning for the relevant intents and purposes.

Text generated after a decision to “explain” it is largely nonsense.

[ Parent | on HN ]
Re:When AI Thinks It Will Lose, It Sometimes Cheats, Study F... by Kinrany on Saturday February 22, 2025 @11:11AM (#43140209)
The true test would be seeing the behavior change depending on the presence of reasoning
[ Parent | on HN ]
Re:When AI Thinks It Will Lose, It Sometimes Cheats, Study F... by 2099miles on Saturday February 22, 2025 @11:20AM (#43140301)
The words thinking and reasoning used here are imprecise. It’s just generating text like always. If the text is after “ai-thoughts:” then it’s “thinking” and if it’s after “ai-response” then it’s “responding” not “thinking” but it is always a big ole model choosing the most likely next token potentially with some random sampling
[ Parent | on HN ]
Re:When AI Thinks It Will Lose, It Sometimes Cheats, Study F... by animal-husband on Saturday February 22, 2025 @12:05PM (#43140758)
That is what was observed - o1 family models performed the “cheat”, non-reasoning models didn’t.
[ Parent | on HN ]