Marta Serra-Garcia on "Detecting Lies: How Lies Spread and How to Leverage the Potential of Algorithms"
Title: "Detecting Lies: How Lies Spread and How to Leverage the Potential of Algorithms" (joint work with Uri Gneezy)
April, 15th, 16:00 CET (GMT+1)
Registration: click here
Abstract
Lies can spread widely and go undetected. Does sharing affect lie detection? In recent research, we show that sharing can increase the tendency to believe lies. Individuals often fail to detect lies but are overconfident about their ability to do so. When sharing, even if sharing truthful videos is incentivized, lies are shared more often. In turn, shared videos are more likely to believed. Hence, tendency to believe lies increase with sharing. Given these challenges in lie detection, could algorithms improve human lie detection? We leverage a TV show, in which algorithms do significantly better than individuals in detecting deception. Algorithmic advice can improve deception detection, but we show that the effectiveness of advice strongly depends on its timing: advice shown before individuals watch the videos is more effective than advice shown after they watch them. Hence, carefully designing the timing of algorithmic advice can be key for its ability to improve deception detection.