Arena Introduces Interaction-as-a-Judge for Agent Evaluation

Arena researcher Wayne Chi, a CMU PhD candidate, introduced Interaction-as-a-Judge, a new evaluation paradigm where computer-use agents interact directly with games and applications to serve as judges, improving agent evaluation.

2026-11-13 ~ 2026-11-13 · 2 related posts

• Arena's Interaction-as-a-Judge: evaluating computer-use agents by how they play — arena · 2026-11-13

1 near-duplicate retellings: arena