Interested parties are welcome to attend.
Presentation Title: Improving Human Deception Detection Using Algorithmic Feedback
(with Uri Gneezy)
Abstract:
Can algorithms help people predict behavior in high-stakes prisoner's dilemmas?
Participants watching the pre-play communication of contestants in the TV show
Golden Balls display a limited ability to predict contestants' behavior, while algorithms
do signicantly better. We provide participants algorithmic advice by agging
videos for which an algorithm predicts a high likelihood of cooperation or defection.
We nd that the eectiveness of ags depends on their timing: participants rely significantly
more on ags shown before they watch the videos than ags shown after they
watch them. These ndings show that the timing of algorithmic feedback is key for its
adoption.
Contact Dan Hungerman for information.