CICERO achieved human-level performance in Diplomacy by combining a language model with planning and reinforcement learning while negotiating with human players.

Format Online paper / PDF

Location PART 3 - THE THIRD AXIS CHAPTER 12: DARK INFORMATION Move 78

"CICERO achieved human-level performance in Diplomacy by combining a language model with planning and reinforcement learning while negotiating with human players."

The Science paper reports that CICERO combined language modeling with strategic reasoning and achieved human-level performance across anonymous online Diplomacy games involving natural-language negotiation.

View Source

Source title Human-level play in the game of Diplomacy by combining language models with strategic reasoning

Source authors Meta Fundamental AI Research Diplomacy Team (FAIR); Anton Bakhtin; Noam Brown; Emily Dinan; et al.

Source year 2022

Confidence high

Importance mid