Ethics in AI Research Seminar (Wednesday - Week 4, TT22)

There is a compelling argument, broadly realist in flavour, in support of the development of automated weapons systems, which is that if these systems are as powerful as they promise to be, a nation which chose not to develop them would incur an unacceptable strategic vulnerability against adversaries which did. Further, specific features of AI technology-in particular, its low 'footprint', in contrast with nuclear weapons-mean that it is difficult if not impossible to verify, publicly, any commitment by adversaries to forego developing AWS. However, resistance to AWS remains widespread, on a variety of grounds. Is there a way of reconciling these positions through a 'no-first-use' policy? This talk explores this question. 

Registration details available here.

_{Ethics in AI Research Seminar Convenors: John Tasioulas and Ted Lechterman | Any queries should be directed to aiethics@philosophy.ox.ac.uk}