This episode is from All Things Go, a podcast about the board game Go, also known as Baduk in Korea and Weiqi in China.
In this episode, Travis speaks with KataGo creator David Wu about how he discovered Go, taught himself programming, and developed one of the game’s most influential AI tools. David explains how KataGo learns through neural networks, search, and self-play. He also discusses the system’s efficiency, its weaknesses, its analysis of smaller board sizes, and its work on the famous Igo Hatsuyōron problem 120. The episode closes with TelegraphGo reflecting on the strange habits and self-consciousness that can arise when moving from online Go to playing in person.
Theme music by UNIVERSFIELD & background music by PodcastAC
David Wu Interview
- A blogpost on David’s beginnings with Katago
- KataGo
- AlphaZero
- The most difficult Go problem ever, the Igo Hatsuyōron and the blogpost on it by David
- KataGoBooks which has the ideal openings currently for different-sized boards
Show your support here
Email: [email protected]