Guojun He (The University of Hong Kong) has posted The AI Lemons Problem on SSRN. Here is the abstract:
I study how AI adoption can undermine prediction markets through an endogenous adverse-selection spiral. When AI raises the informational advantage of some traders, market makers widen spreads, pricing out uninformed participants whose exit further concentrates informed order flow. In the model, beyond a critical AI penetration threshold no active equilibrium survives. Under a participation-based welfare criterion, AI adoption can reduce trading surplus and may be privately excessive because adopters ignore the participation externality they impose. Moreover, the collapse destroys not only trading surplus but also the market’s capacity to generate informative public prices—its core social function—so that even a planner who values only price informativeness would find unchecked AI adoption suboptimal. The mechanism extends to imperfect AI, heterogeneous quality, reduced-form dynamics, automated market makers, multi-market environments, and richer distributions of predictive ability.
Recommended.
To receive a daily summary of posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.