When Turing Met Grotius AI, Indeterminism, and Responsibility

Abstract

The exponential advancement of artificial intelligence and robotics have long intrigued militaries around the world; to the extent, they are developing and deploying lethal machines that would identify, decide and attack enemies and enemy objects, without human intervention. Although these technologies are not fully matured, legal scholars, policymakers, and ethicists already wrongfully presume them to be mere weapons. This article argues that such categorization will deem most actions executed by such autonomous entities (AA entities) neither internationally attributable to states nor criminally attributable to their programmers or military commanders, yielding impunity for the gravest of crimes against civilian populations. This article draws from artificial intelligence theories, the Laws of Armed Conflict and International Criminal Law to refute this assumption and argues a new category of entities should be recognized in order to sustain responsibility and accountability to states and individuals. Scholars are offering different standards and degrees of human control over AA entities, which should make their actions attributable to states and individuals; this article suggests these standards ignore peoples’ over-reliance on computers, and could eventually throw commanders and software programmers under the bus. The ongoing debate tries to fit a square peg into a round hole, asking a cutting-edge technology to be governed by rules created in the 1940s and 70s. Facts and common contradictions fleshed out by this article should spark a new, informed, and progressive debate, which should eventually bring to a fitting and progressive solution for these up-and-coming technologies.