Artificial Intelligence

Abstract

Artificial intelligences are machines that can perform tasks that are characteristically thought of as requiring intelligence. This entry distinguishes between four different types of artificial intelligence (AI): domain‐specific AI, artificial general intelligence (AGI), sentient AI, and “superintelligence.” Existing AI, which is domain‐specific, raises concerns about algorithmic bias, privacy, surveillance, and social impacts, including the possibility of mass unemployment. It is also vital that decisions reached by AI are available for public scrutiny and justification; the question of when we might be justified in trusting decisions reached by AI remains open. Military uses of AI are especially controversial. Should AGI be realized, these issues will become even more urgent and will be exacerbated by the possibility that political and economic questions might be handed over to artificial general intelligences. The idea of machine sentience raises the problem of other minds in an especially stark form. Questions would also arise as to whether sentient machines would have moral status or be moral persons. Might they even acquire more moral standing than human beings? The suggestion that AI might lead to the emergence of superintelligences, which might pose a threat to the human species, highlights an issue that is central to the ethics of all these sorts of AI: who has the right to make decisions about technologies that have the potential to radically the world we all share?