Cass Sunstein: belief

Transcript context

…Five years from now, should we have such trials run by AIs? It would be much cheaper, right? I think the idea of AI trials looks very thinkable today. What role for the human overseer is TBD. We’re not there yet, but with respect to certain things, it would be very disagreeable to many, and maybe that’s a reason not to do it. Maybe it’ll be decreasingly disagreeable to many as people see well-functioning AI as in airlines, and increasingly, motor vehicles. This is at least something to be discussed. I work a lot on the topic of algorithm aversion. We have some research on that and also on AI aversion, where we have less research. Some of algorithm aversion, a cousin of AI aversion, is based on just mistakes, but some of it is based either on a belief that, in some context, human beings will know things that algorithms and AI don’t. Maybe that’s decreasingly true, but it’s true in some contexts. It’s also based on a belief that a human, let’s say judge or participant, is deeply sought, even if accuracy isn’t increased by virtue of human engagement. The ostensible right to a jury by one’s peers — does that matter here?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.