AI devices and liability

We experimentally investigate the effect of the incentives provided by different allocations of liability in the case of (semi)autonomous devices which are a source of risk of accident. Considering three key agents, an AI provider (scientist), a producer, and a consumer, we look at the effect of different liability-sharing rules on the decision-making of each type of agent. We show that assigning liability to the scientist and to the producer is effective in reducing their misbehavior. We also find that assigning liability to the consumer increases her incentive to control the risk of accident in the case of semi-autonomous devices. However, the absence of consumer control (fully autonomous device), coupled with the assignment of liability, decreases the consumer’s propensity to buy the good in the first place. We complete our study with a social welfare analysis which highlights the importance of assigning liability to the producer so that the consumer can have greater confidence in the technology, especially in the case of fully autonomous devices.