Abstract

How do we deal with machines that can act but can’t think?

The social revolution created by effective machine learning systems gives rise to a huge number of legal questions, but one largely unrecognized legal issue is intentionality. Intention is a core concept in a large number of legal domains. In copyright this is expressed in requirements of human authorship and liability for intentional infringements; in torts it is shown in the reasonable person test and causation; and criminal law addresses a range of issues of intentionality including mens rea requirements, mitigating and aggravating circumstances in sentencing, among others.

A number of commenters have sought to resolve the question of intention and autonomous systems by reference to analogous historical examples—typically the fiction of legal personhood for corporations or the liability of slaveowners for their slaves under ancient (typically Roman or Ancient Greek) legal systems. But none have sought to sought to provide a general theory of intentionality in the machine learning age, nor have they sought to examine the implications of functional conscious systems and legal intention across numerous areas of law. This presentation will provide a guide on how to deal with intentional machines.