Language models in wargames: control over actions vs. adjudication

A new paper from arXiv explores the role of language models in open-ended wargames, distinguishing between models that control player actions and those that adjudicate consequences. The research indicates that most studies (~91%) do not grant language models full control over both aspects. The authors emphasize that for simulations to be considered faithful, language models must not only generate plausible agent behavior but also reliably function as adjudicators and world models. AI

IMPACT Clarifies the distinct roles language models can play in simulations, impacting how AI-driven wargames are designed and evaluated.

RANK_REASON The cluster contains an academic paper published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Glenn Matlin
• Hugging Face

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →