AI Revolutionizes Professional Go: Players Now Train by Mimicking Machine Moves
Key Takeaways
- ▸Professional Go players now train primarily by studying and replicating AI moves, with top player Shin Jin-seo matching AI recommendations 37.5% of the time
- ▸AI programs like KataGo have overturned centuries-old Go principles and introduced entirely new strategies that players must adopt to compete professionally
- ▸The democratization of AI training tools is increasing gender diversity in professional Go as barriers to elite-level training decrease
Summary
Ten years after AlphaGo's historic victory over Lee Sedol, artificial intelligence has fundamentally transformed how professional Go players train and compete. Players now spend their days studying AI programs like KataGo, attempting to replicate machine-recommended moves rather than developing purely human strategies. Top-ranked player Shin Jin-seo, nicknamed 'Shintelligence,' matches AI moves 37.5% of the time and spends most of his waking hours analyzing the program's decisions, calling it 'almost like an ascetic practice.'
The shift represents a complete overhaul of Go's centuries-old playing principles. Modern AI programs like KataGo, built on open-source models inspired by AlphaGo Zero, have introduced entirely new strategies that players must master to remain competitive. The Korea Baduk Association reports that training rooms once filled with the sound of wooden Go stones now echo with mouse clicks as players huddle around monitors, debating moves and comparing their choices against AI recommendations. Some players believe this has drained creativity from the game, while others see opportunities for continued human innovation.
The AI transformation has also democratized access to elite-level training, with notable effects on gender representation in professional Go. As AI tools become more widely available, more female players are climbing the competitive ranks. The Korea Baduk Association has reportedly reached out to Google DeepMind about arranging a commemorative match between Shin and AlphaGo for the 10th anniversary, though Shin believes he could now defeat the original AlphaGo by targeting its weaknesses with training from more advanced programs.
- Ten years after AlphaGo's victory, it is now essentially impossible to compete professionally without using AI training programs
