Effect of Fake News on Cascading Behavior
https://www.theguardian.com/technology/2018/nov/12/deep-fakes-fake-news-truth
In this article, Oscar Schwartz warns of the potential harms that “deep fakes” can cause. Unlike fake news seen with misinformation in articles and posts in social media, deep fakes are seen in the form of generated videos and photos that were never actually recorded/taken. Because of this additional visual aspect and how it’s relatively new, it’s possible to believe that it has more power in deceiving people into thinking that a person has said or done something that they, in fact, never have. Schwartz provides such an example where a video of Donald Trump, which shows him weighing in to the people of Belgium on climate change, generated a huge reaction from the public demonstrating that deep fakes can cause information cascades and alter behaviors. Schwartz continues the article by citing different sources on the possible effects that deep fakes will have on the future. For example, Danielle Citron, a law professor at the University of Maryland, and Bobby Chesney are increasingly concerned how these deep fakes can skew information in such a way that would cause greater ideological divides that would push parts of a network “deeper into their own subjective realities”. This is especially prevalent with an increasingly divided political climate. Others, however, such as Tim Hwang, director of Harvard-MIT Ethics and Governance of Artificial Intelligence Initiative, thinks that in the immediate future, such as the upcoming presidential election, will not be greatly affected by deep fakes because generating these deep fakes using machine learning requires a much greater effort than already wide-spread forms of Photoshop-manipulated news.
In class, we’ve discussed how the diffusion of a behavior depends on the information spread through a network. Deep fakes can be especially dangerous because of their inherent nature that they are created to generate strong reactions and outrageous content. For this reason, it’s not hard to believe that a particularly well-made deep fake along with the six degrees of separation phenomenon can allow misinformation to spread widely through a network. This can cause information cascades, and hence, behavioral cascades as well. What’s particularly troublesome is that the people in the network now have to decide what information is accurate and what is not. In the textbook, it’s explained that the Apple Macintosh made very good use of the Super Bowl commercials in advertising their product because such a large portion of the network now knows that other people have viewed this information as well. In regards to viral deep fakes, social media has made it easy for users to see how many other users have seen the post and the reactions of other users. Normally, with completely accurate information, it would be easier to see how a behavioral cascade would occur, but now that it’s unclear as to how many people will actually trust the information, additional complexities occur in analyzing behavior in a network. People’s perceived payoffs may be wildly inaccurate with the influence of deep fakes causing it to be incredibly difficult to accurately model behavior and information in a network.