The Deep Faked Death of Reality

SBGB Highlights from The continued influence of AI-generated deepfake videos despite transparency warnings

We also conducted a conditional analysis for the subset of 47 participants who were shown a warning (either generic or specific), followed by the deepfake video, and subsequently indicated that they believed the warning and therefore knew the video to be fake. Mean perception of guilt for this subset of participants who accepted the warning (0.43) was still greater than in the control condition (–0.29), t(77.52) = 2.13, p = 0.036, d = 0.44, 95% CI [0.03, 0.85]. Qualitative responses¹ indicated that, despite believing the warning, 44.7% (n = 21) of this subset of participants nevertheless relied on the content of the video to conclude that John Carter was guilty, compared with 10.4% (n = 5) of 48 participants in the control condition.

In a separate experiment in the same study…

75 people were shown a deepfake video and warned that it was a deepfake video beforehand. 53% of these people believed the video, and only 13.3% of these cited the warning at the beginning of the video as reason for them not believing the deepfake.

1. Participants were also given the opportunity to provide qualitative responses for perception of guilt, “Please explain briefly why you answered [e.g., probably yes] to whether you think John Carter is guilty of a crime”, and perception of fakeness, “Please explain briefly why you answered [e.g., probably not] to whether you think the video is a deepfake”, in their own words. ↩︎