I am fortunate to be a lecturer at Laval University where I teach a course on design and technology. In this course, we cover several topics, including deepfakes. For those who don’t know, this new technology makes it possible to deceive the public by presenting a video or sound recording that impersonates a public figure (typically).
The term comes from “deep”, in reference to deep learning (deep learning), a form of artificial intelligence, and “fake” which means something fake in English. The first appearances of deepfakes consisted of using the image of well-known personalities to alter the faces of pornographic actors and actresses. However, it became clear that the deepfakes represent an immense potential danger when used as a tool of disinformation.
Returning to the lesson, I invited students to think about potential solutions to prevent the spread of deepfakes. The exercise is interesting for their training in design, but also because it is a new problem for them. Their proposals include all kinds of credible and original solutions, whether technological, operational, legislative or educational.
Among the proposed ideas, one of them deserves a special mention: that of confirming the authenticity of the original videos through journalistic sources.
We often tend to turn to technological solutions to identify the deepfakes, but it puts us in a perpetual cycle of racing for technology. The tools used to generate this type of video are constantly improving, making the recognition exercise more and more difficult. Confirming the veracity of a video post by journalistic sources could be an effective solution to prevent potential misinformation.
Although there is a certain crisis of confidence in journalists, we still live in a society where the institution of journalism is an important mechanism of a democracy. The implication of this statement is deeper than it appears, and the phenomenon of deepfakes is a good example.
The human factor
Especially in the context of a fake video featuring President Zelensky asking Ukrainian soldiers to lay down their arms recently emerged, or a fake interview with Prime Minister Justin Trudeau making a surprising appearance on Joe Rogan’s podcast. Confirmation by journalists implies (or should require) physical presence during the filming of said video. A press conference is a good example. This presence makes it possible to authenticate the person speaking while also validating the time and place of the recording, which is difficult to counterfeit.
Moreover, the fact of relying on several journalists, from several different channels, acts as a redundancy factor, reducing the risk of information manipulation. Collusion is not impossible, but more difficult. I understand that, for some, a doubt may remain and that they could be inclined to point to a new world order in which journalists would be accomplices. That’s true, but it’s highly unlikely. In fact, there is no perfect system, but there is something fundamental about trusting our institutions.
Other solutions, such as blockchain authentication, were highlighted. However, there’s something appealing about relying on people, not technology.