Ulric B. and Evelyn L. Bray Social Sciences Seminar

Abstract: Innovation often requires combining complementary knowledge, but sharing ideas risks expropriation. We analyze how decision-makers with incomplete projects search for collaborators through safe small talk, private conversations within their network, or exposing ideas entirely via public broadcasting. Broadcasting is efficient for finding matches but risks expropriation, while private conversations provide more control but are less efficient. "AI enclaves" can match decision-makers based on private information about complementarities in their ideas without revealing the ideas themselves, providing a novel application of agentic AI. We investigate to what extent such enclaves can help overcome search frictions and encourage information sharing.