Deepfakes are warping reality. This AI project turns them into a history lesson
Briefly

Deepfakes are warping reality. This AI project turns them into a history lesson
""The Great Dictator," which premiered this week in Austin, flips the script on what deepfakes have come to represent. Instead of using generative AI to create misinformation, it uses AI video and voice tools to blend participants into archival footage to experience history through their own voice and likeness."
""This is an exhibit that examines something that was as powerful 3,000 years ago with no technology, with the ancient Greeks," Arora says. "It really shows you we might have all the technology we want, and humans don't change. We have something hardwired in us about rhetoric and power and someone speaking up.""
"At a hotel in downtown Austin, attendees step up to a podium flanked by three large screens cycling archival footage. After consenting for their voice and likeness to be used, the person then chooses one of three speeches from three very different eras: Malcolm X's 1964 "The Ballot or the Bullet" speech in Cleveland, Ronald Reagan's 1987 "Tear down this wall" speech in Germany, and Zohran Mamdani's 2024 victory speech in New York City."
"The Great Dictator" is an interactive art installation debuting at SXSW that reimagines deepfake technology for positive purposes. Participants record themselves delivering excerpts from three historically significant speeches: Malcolm X's 1964 "The Ballot or the Bullet," Ronald Reagan's 1987 "Tear down this wall," and Zohran Mamdani's 2024 victory speech. Using AI voice and video tools, the installation seamlessly inserts participants' cloned voices and likenesses into archival footage while an AI-generated crowd reacts authentically to their delivery. Created by filmmaker Gabo Arora, the project explores how emerging technology can serve educational and artistic goals rather than profit, warfare, or propaganda, while examining timeless human elements of rhetoric and power.
Read at Fast Company
Unable to calculate read time
[
|
]