The race is on to produce artificial general intelligence (AGI)—machines that are at least as intelligent as humans—despite widespread concern that an AGI would pose an existential threat to humankind.
The fundamental problem is that much research on AGI is premised on the mistaken assumption that for a machine to be as intelligent as a human requires no more than that it produces intelligent behaviour. So, what is most likely to happen in the near future is that we will create an ersatz-AGI: a machine that is capable of mimicking intelligent behaviour, without having the capacities constitutive of human intelligence—most notably, without any capacity for moral action.
In order to develop a true AGI, we need to revisit philosophical questions about the nature and requirements of distinctively human intelligence. The trouble is that the existing philosophical models of human cognition have major shortcomings. I sketch a novel account of the nature of distinctively human intelligence, and a blueprint for the construction of a true and moral AGI: a machine with the cognitive capacities constitutive of human intelligence, including the capacity for moral action.
Research seminar with Anandi Hattiangadi, professor of philosophy at Stockholm University and a researcher at the Institute for Futures Studies. She is currently an affiliated researcher at the Institute of Philosophy, University of London, where she is a participant in the AI and Humanity Project.