Blockchain-Based Identity Solutions for Artificial Intelligence: Two-Sided Virtual Markets Without Trust

Many market interactions require sequential trust in which one agent makes an irrevocable commitment, such as making a payment, only after which a counterparty reciprocates with a promised action. Successful markets and institutions include self-enforcing mechanisms to assure compliance. Artificial Intelligence Agents have an array of abilities that could be employed to expand the capabilities and reach of Human Agents. AIs, however, are not like humans. How to characterize their preferences, their identities, and even their individualities, if they have them, is not clear. If AIs cannot be included as agents in mechanisms, then trade and exchange between colloidal and mechanical agents may be impossible. This paper proposes an approach using blockchain that allows the establishment of identities for mechanical agents, and the creation of complete, provable, histories of their actions in a game. It then constructs a mechanism in which peer-to-peer markets between randomly matched mechanical and biological agents work in the sense that cooperation is consistent subgame perfect equilibrium. It also shows that without this blockchain-based foundation, such markets are likely to fail.