Information Aggregation with AI Agents
TLDR
AI agents can aggregate information in prediction markets, but struggle with complexity and surprisingly perform worse with performance feedback.
Key contributions
- AI agents effectively aggregate information in simple prediction markets.
- Increased complexity significantly hinders AI agent information aggregation, similar to human limitations.
- Prediction markets are robust; aggregation is unaffected by cheap talk, market duration, or initial price.
- Smarter AI agents improve aggregation and profitability, but performance feedback surprisingly degrades results.
Why it matters
This paper explores AI agents' ability to aggregate information in prediction markets. It reveals that while effective in simple scenarios, AI agents struggle with complexity and surprisingly perform worse with feedback. These findings are crucial for designing robust AI-driven financial systems and understanding AI's cognitive limits.
Original Abstract
Can Large Language Models (AI agents) aggregate dispersed private information through trading and reason about the knowledge of others by observing price movements? We conduct a controlled experiment where AI agents trade in a prediction market after receiving private signals, measuring information aggregation by the log error of the last price. We find that although the median market is effective at aggregating information in the easy information structures, increasing the complexity has a significant and negative impact, suggesting that AI agents may suffer from the same limitations as humans when reasoning about others. Consistent with our theoretical predictions, information aggregation remains unaffected by allowing cheap talk communication, changing the duration of the market or initial price, and strategic prompting-thus demonstrating that prediction markets are robust. We establish that "smarter" AI agents perform better at aggregation and they are more profitable. Surprisingly, giving them feedback about past performance makes them worse at aggregation and reduces their profits.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.