In an era of rapidly advancing technology, the rise of artificial intelligence (AI) has presented humanity with both incredible opportunities and daunting challenges. Among these challenges, one of the most concerning is the growing prevalence of AI-generated disinformation. As AI algorithms become increasingly sophisticated, the potential for malicious actors to exploit this technology for spreading misinformation and manipulating public opinion becomes a pressing concern.
When Donald Trump was about to be indicted in March, Bellingcat founder- Eliot Higgins, decided to go to Midjourney- the image-generating program. Prompts like “Donald Trump falling on the ground while getting arrested” led to images that look genuinely real to anyone looking at these photos passively.
These photos were spread even further than they should have because of Higgin’s social media reach, meaning the retweets and reposts of the deep fakes pushed them outside of his following and onto people who don’t know better. I was one of the people who were outside of Higgin’s circle of followers, so the legitimacy of these images was uncertain at best, especially when people just shared the image without the context of their AI origins.
The image generation tools like Midjourney and language algorithms like ChatGPT are still in their infancy in terms of their proficiency and how well they mimic human quirks. ChatGPT rarely, if at all, makes grammatical errors or comes up with sentences that go nowhere. Even so, their effects on information sharing are astronomical.
There is a certain genre of internet memes that is very popular on short-form video content-sharing apps like Tik-Tok, Instagram Reels, and Youtube Shorts. These memes portray Presidents of the United States of America like Barack Obama, Donald Trump, and Joe Biden, and sometimes other influential figures in the online spaces like singer/rapper Ice Spice, podcaster Joe Rogan, right-wing political pundit Ben Shapiro, and Canadian philosopher Jordan Peterson playing video games or telling jokes or acting like a group of friends, all of which is deep faked using many AI algorithms, even the voices that sound eerily similar to their real voices.
These videos are fake. Nobody of sound mind will think that all of these public figures hang out with each other playing video games and making fun of each other like they are teenagers. However, this gives us a taste of what can be done with these tools right now, and what could be done with newer and better tools very soon.
Back when AI image generators were new, many who were mad at the programs taking jobs from real artists liked to joke about the AI’s inability to understand how human hands work. It has been a few months since these arguments gained popularity, and AI has evolved yet again, producing beautiful hands that look indistinguishable from artist creations.
Deep fakes in some form have always existed, even before ChatGPT and Midjourney, but they had to be made by people who knew how to use the software and required several painstaking hours. Technology that can make the production of deep fakes of even the current president of the most influential country is scary.
Just in the spirit of things, the opening paragraph was written entirely by ChatGPT.