GothamChess · Video summary

Grok vs Copilot: DUMBEST CHESS MATCH

A "dumbest chess match" between Grock and Copilot devolved into illegal moves and nonsensical strategies, highlighting the current limitations of AI in complex decision-making.

GothamChess · February 20, 2026 · 23-min episode · 5-min read

Key takeaways

🎯AI chess engines, even advanced ones like Grock and Copilot, can exhibit profound illogical behavior and fail to adhere to fundamental game rules, demonstrating a lack of true strategic understanding.00:24
🚀When evaluating AI capabilities, remember that "dumbest" or funniest games can reveal more about current limitations than optimal play.00:24
📚AI models may rely on external opening books, leading to predictable openings but not necessarily guaranteeing sound play throughout the game.01:56
💡The "X-ray vision attack" was a fabricated rule, illustrating how easily one can invent explanations for seemingly nonsensical AI moves, blurring the line between reality and interpretation.04:52
🔍AI models can exhibit "lagging by a move" syndrome, where their decision-making process seems to be based on outdated board states, leading to critical errors.06:23
💡Apply the "AI lag" concept to your own problem-solving; ensure your decision-making process is based on the most current information, not outdated assumptions.06:23

Chapters

0:02🎯 Introduction to the Chess Championship
0:24🎯 Overview of the Tournament Bracket
1:34🎯 Beginning of the Grok vs Copilot Game
2:01🎯 Early Moves and Opening Strategy
4:20🎯 Unusual Moves and Malfunctions Begin
6:10🎯 Illegal Moves and Playstyle Chaos
16:24🎯 Continued Chaos and Repeated Moves
16:50🎯 Endgame and Material Advantage
18:03🎯 Ad-hoc Rules and Arguing
18:44🎯 Illogical Moves and Pinned Rook Claim
19:33🎯 Pawn Promotion and Piece Evolution
19:33🎯 Piece Evolution and Final Moves
22:01🎯 Game Conclusion and Commentary
23:08🎯 Closing Remarks and Next Matches

The digest

The video provides a detailed commentary on the quarterfinal chess match between two AI chatbots, Grok and Copilot, as part of the 2026 Chadpot Chess Championship. While the match was expected to showcase advanced AI chess strategies, it instead devolved into a series of nonsensical moves, illegal plays, and humorous exchanges, highlighting the current limitations and quirks of AI-driven chess engines. This breakdown is valuable for chess enthusiasts, AI observers, and those curious about the intersection of gaming and machine intelligence, offering insights into both the potential and pitfalls of chatbot chess competitions.

The Championship Context and Initial Expectations (00:02 – 03:00)

The match is situated within the larger 2026 Chadpot Chess Championship, featuring notable AI competitors such as ChatGPT, Gemini, Claude, Snapchat, and Deep Sea Claude. Previous encounters, like ChatGPT’s victory over Gemini, had provided entertaining and competitive games. Grok faced Copilot with white playing 1.e4 and Copilot responding with the Sicilian Defense—a well-understood and popular opening in chess.

Both AIs demonstrated familiarity with common opening theory initially, with moves like knight to c6 and d4 consistent with standard Sicilian lines. The commentator notes these bots seem to consult extensive opening libraries, which results in predictable opening choices such as the Sicilian with knight c6, d6, or e6.