Data Ethics & Society Reading Group 03-12-24, 13:30 GMT
You're welcome to join us for our next Data Ethics & Society Reading Group on DAY the DATE at 13:30 GMT. This time we're going to read The Alignment Problem by Brian Christian, which got the #1 spot on The New York Times' Five Best Books About Artificial Intelligence and which "Masterfully surveys the ‘AI fairness’ community, introducing us to some of its main characters, some of its historical roots, and crucially, many of its philosophical quandaries and limitations" according to Cathy O’Neil, author of Weapons of Math Destruction, which we read in May 2023.
Thank you to everyone who voted for this content.
There will be time to talk about whatever we like, relating to the book, but here are some specific questions to think about while you're reading.
- Both humans and AI systems are prone to bias. If we accept that some degree of bias is inevitable in AI, how do we determine what level of bias is acceptable in different contexts?
- The book explores the contrast between simple, interpretable models and more complex, less transparent ones in machine learning. It raises questions such as: When should we prioritise transparency over accuracy, and when is accuracy more important? Under what circumstances is a simple model the most suitable choice?