Shu Hu, a Â鶹¾«Æ· assistant professor of computer and information technology in Indianapolis, recently sat down with the Associated Press (AP) newsroom to on deepfakes.
Hu is the recipient of a National Science Foundation Pilot award, which grants him and graduate students the opportunity to use the Department of Energy’s Lonestar supercomputer for investigating AI-powered techniques for detecting deepfakes.
As this newsroom wrote at the time of Hu’s award: “The term ‘deepfake’ refers to video content that has been digitally manipulated to make it appear as though someone is present, or even speaking in a video, when their appearance has in fact been fabricated using a combination of graphical technology and artificial intelligence.”
The AP summarized contents from the video featuring Hu as such:
“Hu says audio manipulation in deepfakes creates the need to accelerate the development of multimode deepfake detectors, especially since he expects deepfakes to become increasingly common. AI is becoming a key tool to help identify deepfakes. According to Hu, developing AI to remember the patterns and cues of deepfake recognition is the next step in deploying AI models in real-world scenarios. He explains how people can use cues such as pupil shape to help identify deepfakes. The biggest challenge in the process of deepfake detection is educating more of the public so they can protect themselves from deepfakes with these methods.”
Watch the original video, with credit to Hu and the AP, .