The Mainichi Shimbun answers some common questions readers may have about "deepfakes," false but convincing digital images and videos.
Question: How do you create deepfake images and videos?
Answer: Deepfakes are created by artificial neural networks (ANN) -- a type of artificial intelligence that imitates the human brain -- that have been trained on enormous quantities of data. And the technology has advanced by setting one AI that creates deepfakes such as of human faces against another AI that detects the fakes.
Q: What kind of things can you make using the technology?
A: In addition to images and videos, it is now possible to create audio that sounds exactly like a specific person. In some cases, deepfake technology has been used to make faces of non-existent people to be used in advertisements. Technology to seamlessly change one's gender, age, skin color and other traits is also advancing.
Q: Wouldn't it be scary if your face and voice were being used without your knowledge?
A: In California, it is now illegal to deceive voters by creating and spreading deepfakes of politicians. There is a risk that people's photos on social media could be used to make deepfakes without their permission. If you find something suspicious about a video or image, you should not share it.
Q: Wouldn't it be a great technology if it wasn't used with malicious intent?
A: There are growing expectations for AI-based technologies in the field of arts and culture. If deepfake technology is used for things like entertainment and painting, it would likely open up a large new market segment. For example, new pieces have been created by having AIs learn the works of Japanese manga artist Osamu Tezuka and paintings by 17th century Dutch master Rembrandt. However, there are still many issues, including how to treat the dignity of the deceased, and deciding who owns the copyright. It is essential to discuss what rules to make.
(Japanese original by Ryo Watanabe, Science & Environment News Department)