After AI beat them, professional Go players got better and more creative

AI as Skill Amplifier in Go and Other Games

• Go AIs shattered long‑standing human “dogma,” revealing many accepted joseki and strategies as suboptimal and legitimizing once‑discouraged moves (e.g., early 3‑3 invasions, kick joseki).
• Pros now train extensively with strong engines (KataGo, LeelaZero, etc.), and aggregate “decision quality” vs AI evaluations shows large improvement; weaker current pros reportedly surpass pre‑AI greats by that metric.
• Commenters stress that improvement isn’t just rote copying: many gains come from “human” moves that differ from AI but are informed by its ideas.
• Similar patterns are noted in chess and Scrabble: engines expose mistakes, remove myths, and raise the average level of serious players.

Dogma, Creativity, and Style Changes

• AI is seen as a dogma‑destroyer, pushing players out of local optima and expanding the explored strategy space.
• Some worry that for amateurs, play gets less “creative” as they mimic AI lines or memorize joseki without understanding. Others counter that real games diverge quickly and blind memorization fails without judgment.
• In chess, some say top‑level play became more defensive and draw‑oriented; others argue that incentives, ratings, and tournament economics, not engines, drive drawish play.

Tools and Ecosystem

• Popular Go tools: KataGo (often considered stronger than AlphaGo), LeelaZero, KaTrain as a frontend.
• Playing servers mentioned: online-go.com (OGS), KGS, Fox, Tygem, goQuest (especially for 9x9).
• Broadcasts now routinely show AI win‑probability overlays, which some find insightful and others feel spoils the “magic.”

Broader Implications for AI and Human Work

• Many see Go as foreshadowing AI augmenting experts in programming, science, and other fields, with specialists using AI as a powerful sparring partner.
• Concern: if AI replaces beginner‑level work, novices might lose key experiences needed to develop independent judgment.
• Several argue that in games and entertainment, humans will remain central as performers, even when AI is vastly stronger, analogous to cars vs sprinters.
• Others expect large parts of film, illustration, music, and vlogging to be AI‑generated, with human art surviving where narrative, identity, or “no‑AI” authenticity is valued.

Limits and Skepticism

• Some note that success in finite games doesn’t equate to general intelligence; engines remain brittle to rule changes and lack broad reasoning.
• “Better” in Go is precisely defined via win probability, but commenters doubt that such clear metrics exist for most real‑world creative domains.