Are you smarter than a deepfake? A psychological perspective on human deepfake detection

• In Event: Cybercrime, Cybersecurity, and Education

Abstract

Disinformation is the dissemination of inaccurate information with the purpose of deceit, typically in the form of fake news or propaganda. One method of spreading disinformation that is becoming increasingly difficult to detect is through an artificial intelligence (AI) technique, deepfake technology. It can be used to doctor images, videos, and audio, in which a person in an existing media form can be replaced by another person. There is also technology that allows AI-based computer programs to generate realistic-looking faces of people who do not exist. These technologies may be used to spread disinformation to sway and influence people’s opinions, whether that is to support or oppose specific views. There is currently a large focus on algorithmic detection of deepfakes, but much less so on human detection. This talk provides the main takeaways from a study on the perception and detection abilities of humans in detecting media that have been manipulated using deepfake technology. This presentation discusses various criteria that people use to determine if a media is genuine or manipulated, along with criteria that create misconceptions in detection. Further the impact of recognition, confidence, and other individual characteristics/demographics on detection will be discussed.

Author

• Rachel Bleiman, Temple University