Xinyu Wang and Andres Fielbaum simulated drivers as self-interested large language model (LLM) agents, using Claude Haiku 5.5 and Jev, to compare reactive and anticipatory rebalancing policies once rules are public.[1] The paper is posted as a preprint on arXiv.org.[1]
Study casts Claude Haiku 5.5 and Jev as self-interested drivers
Xinyu Wang and Andres Fielbaum compare reactive and anticipatory rebalancing policies once the rules are public, in a paper posted as a preprint.
1 primary · 0 independent · 0 commentary · 0 vendor