In this age of fake news spreading virally over social media, the recent Harvard Neiman Lab article (link below) is an excellent read.
As I am quoted in the article, “The way to combat deepfakes is to augment humans with artificial intelligence tools.” Humans alone, or technologies alone are ineffective at defeating this growing problem. However, humans augmented with artificial intelligence (AI) technology can be a formidable defense against fake news.
Imagine a fake video clip or photograph that shows a person doing something they didn’t. AI alone may not be able to detect that the video or photo is fake, but AI combined with a human detective — an investigative journalist — can research and examine real world knowledge and information beyond the reach of the AI and determine that it is fake.
The investigative journalist could uncover and confirm from other sources that the person shown in the video or photo couldn’t have been in it because of contradictory real world information. For example, the subject may not have been alive when the video claims to be shot, or have been of a different age, or looked remarkably different in other videos and photos from the same time period.
If you are interested in learning about this fascinating topic — how deep learning is used to both create and combat fake news — I highly recommend reading the article about how The Wall Street Journal is preparing its journalists to detect deepfakes by my colleagues Francesco Marconi and Till Daldrup. The article is approachable, engaging, and educational. It also contains some enjoyable videos and a fun quiz.