{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T03:29:06Z","timestamp":1773804546364,"version":"3.50.1"},"reference-count":0,"publisher":"Association for the Advancement of Artificial Intelligence (AAAI)","issue":"35","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["AAAI"],"abstract":"<jats:p>Economic decision\u2011making depends not only on structured signals\u2014such as prices and taxes\u2014but also on unstructured language, including peer dialogue and media narratives. While multi\u2011agent reinforcement learning (MARL) has shown promise in optimizing economic decisions, it struggles with the semantic ambiguity and contextual richness of language. We propose LAMP (Language\u2011Augmented Multi\u2011Agent Policy), the first framework to integrate language into economic decision\u2011making, narrowing the gap to real\u2011world settings.\nLAMP follows a Think\u2013Speak\u2013Decide pipeline:\n(1) Think interprets numerical observations to extract short\u2011term shocks and long\u2011term trends, caching high\u2011value reasoning trajectories.\n(2) Speak crafts and exchanges strategic messages based on the reasoning, updating beliefs by parsing peer communications.\n(3) Decide fuses numerical data, reasoning, and reflections into a MARL policy to optimize language\u2011augmented decision\u2011making.\nExperiments in economic simulation show that LAMP outperforms both MARL and LLM\u2011only baselines in cumulative return (+63.5%, +34.0%), robustness (+18.8%, +59.4%), and interpretability. These results demonstrate the potential of language\u2011augmented policies to deliver more effective and robust economic strategies.<\/jats:p>","DOI":"10.1609\/aaai.v40i35.40201","type":"journal-article","created":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T02:33:06Z","timestamp":1773801186000},"page":"29582-29590","source":"Crossref","is-referenced-by-count":0,"title":["Think, Speak, Decide: Language-Augmented Multi-Agent Reinforcement Learning for Economic Decision-Making"],"prefix":"10.1609","volume":"40","author":[{"given":"Heyang","family":"Ma","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qirui","family":"Mi","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qipeng","family":"Yang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zijun","family":"Fan","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Bo","family":"Li","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Haifeng","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"9382","published-online":{"date-parts":[[2026,3,14]]},"container-title":["Proceedings of the AAAI Conference on Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/40201\/44162","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/40201\/44162","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,18]],"date-time":"2026-03-18T02:33:06Z","timestamp":1773801186000},"score":1,"resource":{"primary":{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/view\/40201"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,14]]},"references-count":0,"journal-issue":{"issue":"35","published-online":{"date-parts":[[2026,3,17]]}},"URL":"https:\/\/doi.org\/10.1609\/aaai.v40i35.40201","relation":{},"ISSN":["2374-3468","2159-5399"],"issn-type":[{"value":"2374-3468","type":"electronic"},{"value":"2159-5399","type":"print"}],"subject":[],"published":{"date-parts":[[2026,3,14]]}}}