Detecting Deepfakes and AI Fakes

#16 - Responsible AI and Building Trust in AI - Liu Feng-Yuan

Topic 6 of 16

Published Nov 23, 2020 · Clip 23:42–25:10 (1:28)

Feng-Yuan explains that AI models can only learn within the boundaries of their training data, so throwing left-field questions at chatbots or generated content reveals whether it's a machine. He notes systematic deepfake detection is an ongoing adversarial arms race akin to cybersecurity, and distinguishing real from fake is a hard, important problem with no easy answers.