Talk from
Alexei Grinbaum – CEA-Saclay/LARSIM
Short abstract:
I’ll present recent work on AI ethics in healthcare and human cognition. In the iRECS project, we train researchers to implement AI ethics “by design”, and I’ll discuss the strengths and weaknesses of this approach. One particularly interesting case study is the use of chat bots in psychiatry. In the AIOLIA project, we’ll work on guidelines for AI ethics in high-risk applications including subliminal influence based on behavioral and brain data. Emergent capacities in self-supervised learning may open up previously unknown possibilities of using such data. The future CCNE opinion and the future EU policy under the AI Act will make use of our recommendations, and I’ll describe several opens questions where contributions are expected from neuroscientists.