Discussion: View Thread

Dear All,

Please join us online on Wednesday, 18 May, at 4pm BST, for the second R:ETRO seminar of this term, hosted by the Oxford University Centre for Corporate Reputation.

Tae Wan Kim (Carnegie Mellon University) will be giving a paper entitled "When is it ethically admissible for artificial intelligence to lie?"

Abstract:

I discuss how to develop conversational AI to behave ethically. I explain why the ethical status of conversational AI in part depends upon whether there is what we call an "invitation to trust" in a particular scenario. Second, I identify scenarios where expectations regarding whether an AI negotiator invites or does not invite trust diverges from expectations for a human negotiator. These expectations subsequently in part determine when it is permissible for AI versus humans to lie, and differing expectations suggest that there may be some contexts where humans are permitted to lie-but not AI. A practical implication is that companies that employ AI in communicating with customers should be more careful and stricter when developing a machine that can utter false statements. The goal of this project is not to develop an expert machine on lying, but a machine that aligns with ethical values in a normatively justified and empirically grounded manner.

Click here to register.

I hope to see many of you there!

All the best,
Rita

------------------------------
Rita Mota
Research Fellow
Saïd Business School, University of Oxford
OXFORD
------------------------------