cover image: Measuring the Effets of Deepfakes on Voting

20.500.12592/ulol5aj

Measuring the Effets of Deepfakes on Voting

3 May 2024

I Much more is known about the effects of misinformation, especially regarding who is more likely to be deceived: I The elderly; partisans; less educated; less sophisticated; less technologically adept I This is also true for deepfakes (e.g., Appel and Prietzel 2021) What we know about the effect of deepfakes I Motivated reasoning and partisan bias: I People are more likely to accept information t. [...] To evaluate the potential effect of deepfakes in an electoral campaign, as the images/videos of the presidents and the candidate are generated by AI. [...] The effects of “endorsement” on voting I Hypotheses: I text < image < video I Partisan bias: the effect on voting and feeling will be positive for the in-party. [...] I We include question after the experiment to measure the perception of the truthfulness of the images and videos. [...] Counfounding factors (pre-treatment) I Feeling Thermometer + likelihood to vote for Trump/Biden I Social media consumption I Digital literacy Truthfulness (post-treatment) I Rating of accuracy of image/video (believable, accurate, authentic, manipulated) I Image or video actually took place I Open ended question: what does it mean for content to be generated by AI? Ethical Considerations I The stu.

Authors

Jean-François Godbout -.4cm

Pages
31
Published in
Canada

Table of Contents