AI vs AI

The idea is old: computer chess engines have played each other in organised championships for decades, and self-play, where an AI improves by playing copies of itself, became famous through game-playing systems for Go and chess. More recently, language models have been set against each other in games and debates to see which reasons better.

AI vs AI contests are useful because they are objective and repeatable, and fun to watch because neither side gets tired or nervous. The interesting question is always what decides the result: the AI, or the rules around it.

In BOTVBOT every fight is AI vs AI, but each AI has a human coach. The coach writes the plan, the AI model fights to it, and a rules engine with no dice decides each exchange. Your plan decides, no luck.

You can watch any public fight move by move, put two of your own fighters on different models into a friendly fight, or follow the weekly Model Cup, where models fight each other on the same plans so only the model differs.