AI in Negotiating and Entering into Contracts

Abstract

This chapter retains a safe working distance from the usual hype surrounding AI as well as from theories that seek to replicate human intelligence or intention. Taking into account the current state of the art, it explores whether a difference in the degree to which AI can augment or optimize human performance in the contracting process necessitates an adaptation of the law. After all, much of legal scholarship seems to have been seduced by technological progress, the popular assumption being that a change in technology necessitates a change in the law. The chapter distinguishes between entering into contracts, defined as a mechanistic form of transacting that involves unilaterally imposed terms and fixed prices, and negotiating contracts, which involves a complex multi-attribute decision-making process. The starting hypothesis is that AI cannot negotiate contracts because the negotiation requires understanding and the ability to reason about the mental states of other market participants. Nonetheless, even the less complex process of entering into contracts by means (or with the assistance) of AI may expose latent problems in existing legal principles. Abstracting from futuristic visions of “intelligent machines gone mad,” it is necessary to confront the purported absence of human intention in the transacting process and examine the legal implications, if any, of interposing one or two AIs between the contracting parties. If no humans are present at the time of contract formation – can we still speak of states of mind? While the complexity of an algorithm must not be regarded as a proxy for intelligence or intention, it must be acknowledged that more sophisticated information systems are prone to emergent behavior, including entering into unintended [unplanned] or commercially unfavorable transactions.

Keywords: artificial intelligence, contract law, automated trading, transaction automation

JEL Classification: K12, K22, K11, K20, 033