The Dead Internet Theory is the belief that the vast majority of internet traffic, posts, and users have been replaced by bots and AI-generated content, thus surpassing the human-run Internet of yesteryear. While it began as a conspiracy theory on controversial message board 4Chan, a more rational version of the idea has entered the mainstream via coverage from major media outlets.
At Reality Defender, we believe that the human Internet is still alive and somewhat well, but the algorithmic trend toward its death is hard to dismiss. The battle royale over the coveted top position on Google’s search results leads websites and creators to play the bot games and focus on gaming the algorithm over content. This is no problem, since content can be an afterthought written by an AI chatbot. Perhaps this wouldn’t be so bad if synthetic content was still read by humans, but research shows that half of all Internet traffic now comes from bots doing anything from scraping websites, learning to emulate human writing (by reading AI-generated content), or attempting to commit ad fraud. Already, too much of the still-alive-Internet can be distilled into bots interacting with bots and AI-generated content flooding the gates.
The Deepfake as Entertainment (and Downfall)
The benefits of the remarkable achievements in AI cannot be dismissed: generative AI and deepfake technology can be utilized in many ways to improve society — making education more accessible and individualized, or helping train the next generation of doctors. The simple pleasure of harmless deepfakes can’t be discounted, either. Who doesn’t enjoy the occasional pope in Balenciaga?
The problem is that the Internet has become a firm pillar of our reality. It is entrenched in our social lives, our economies, it runs our jobs searches and dispenses our news and tells us where we can vote, and thus, its descent into the dead, the automated, the unreal, a swirl of AI-generated articles and images and bots creates a perilous environment for our already-fractured sense of what is and isn’t real.
Yet we believe that the Internet’s actual death isn’t certain. Keeping the internet alive even as it’s overrun with fakes is possible as long as we deploy robust systems to keep track of and properly label all AI-generated content. To start, we can combine powerful inference detection tools that scan content to look for telltale signs of AI manipulation and provenance tools that encode the content’s synthetic nature within its metadata. This approach, if adopted at scale by companies and institutions willing to work with each other, creates a comprehensive method to immediately alert users that the content they’re viewing has a complicated relationship with reality.
This is one of the simplest and yet most effective ways of drawing a line between the human Internet and the Internet of bots-for-bots—a line that must be drawn sooner than later. A clear understanding of whether what we are seeing, hearing, and experiencing is an AI creation will allow humans to navigate the inevitably artificial version of the Internet with vigilance, or enable us to opt out of it altogether, should we choose to. To remove this choice from users is to take a distinctly human invention—one based upon our need to know each other, to communicate outside the boundaries of space and time—and turn it into a storm of clashing synthesized chaos whose only effect is to deaden, divide, and dehumanize us.
