Credit-Blame Asymmetries in Human Use of Personalised versus ‘Standard’ Generative AI
As generative AI becomes more thoroughly integrated into many human activities, we may face a systematic credit-blame asymmetry: We may not deserve full credit or praise for the valuable outputs we create with generative AI (i.e., if we do not put in sufficient skill or effort), but we may be entirely blameworthy for harmful outputs (e.g., due to negligence or recklessness). How might patterns of praise and blame change, however, if we use personalised AI that is trained on our own past outputs, created solely by us? In this talk I present recent theory and evidence on this question with data from the US, UK, China, and Singapore.
Date: 11 October 2023, 12:30 (Wednesday, 1st week, Michaelmas 2023)
Venue: Please register to receive venue details
Speaker: Dr Brian Earp (University of Oxford)
Organiser contact email address: aiethics@philosophy.ox.ac.uk
Host: Professor John Tasioulas (University of Oxford)
Part of: Ethics in AI Lunchtime Seminars
Booking required?: Required
Booking url: https://www.oxford-aiethics.ox.ac.uk/ethics-ai-lunchtime-research-seminars-credit-blame-asymmetries-human-use-personalised-versus
Booking email: aiethics@philosophy.ox.ac.uk
Cost: Free
Audience: Public
Editors: Marie Watson, Lauren Czerniawska