Bruno Marnette comments on Modeling incentives at scale using LLMs