Search + Learning + Self-Play, this triangular structure is indeed classic. Now large models have somewhat lost the aspect of search.

AIMPACT News, May 16 (UTC+8), using AlphaGo as an example, explains the fundamental components of intelligence. AlphaGo remains the clearest and most comprehensive example to date, integrating three core fundamentals: search techniques, learning from experience, and self-play. These three elements together form the key pathway to surpass human chess skills. (Source: AiHot)