Identifying and quantifying human and AI agent collaboration conventions in Hanabi
What happened
A paper on arXiv proposes the "convention gap" metric to compare how humans and layered AI agents use cooperative conventions in the card game Hanabi. According to the paper, across two independently collected sets of human games, humans have a larger convention gap than every agent tier. However, on conventions both sides use, humans only match Tier 2 agents. Humans' advantage appears mainly in situations where a clue points to a card that has not yet been touched. The paper also reports that convention use relates to players' knowledge of what their cards are, and that it appears mostly in the mid-game. Masking several high-gap situations at once has a much larger effect than the sum of masking each one individually.
AI-written from the coverage · updated 2026-10-15
Coverage timeline
Follow the coverage to see the story from different sides.
- arXiv Board & Classic Game AIPickWhat and where in identifying and quantifying conventions in humans and AI agents
The paper proposes a convention gap metric in the card game Hanabi to compare how humans and layered agents use conventions. In two independently collected sets of human games, humans have a larger gap than every agent tier. On conventions both sides share, however, humans only match Tier 2, and their advantage is concentrated in clues that point to untouched cards. Convention use relates to players' knowledge of their cards and is most common in the mid-game. Masking several high-gap situations has a much larger effect than the sum of masking each one alone.