The Matrix is quickly becoming real. As reality slides slowly into chaos, scientists are intent on creating computer models of it to train AI systems in the quest for General Artificial Intelligence (AGI). Nvidia, for instance, is building a “multiverse simulation engine” to provide the massive amounts of data required to train generative AI systems to do anything. These neural nets would produce huge amounts of realistic videos based on real-world physics dealing with all potential outcomes of situations to train self-driving cars, among many other things. These would use statistical algorithms to decide each step of the simulation and apply same magical diffusion technology as generative AI platforms like Midjourney or Stable Diffusion to create the images.

‘Multiverse simulation engine’ predicts every possible future to train humanoid robots and self-driving cars

Meanwhile, Google is also staffing up in its effort to create real-world simulations with generative AI. It sounds like they wish to train robots and online avatars and create holodeck-like virtual environments in real time for entertainment. To do this requires a world model (like the ones we carry around in our head) that predicts physical consequences of actions as well as the generative capability to depict them, just like the Nvidia project.

Google Wants to Simulate the World With AI

So what could possibly go wrong? Well, there’s the whole Matrix thing, to start with. Secondly, AI systems could finish hoovering up all publicly-available human knowledge on the internet by as early as next year, and then what would they do? Plumb the secret depths of the net and salvage private data squirreled away, like that on Facebook? They’ve already gotten into trouble with bad advice from Reddit and The Onion that the machines took seriously.

Or, as these plans reveal, make fake information? The problem with that is that regurgitated AI content often quickly turns to crap. This can lead to not only “model collapse” but fill the net with “unintelligible gibberish” — kind of like a Kessler syndrome in cyberspace.

And it already may be happening. The “dead internet theory” says that the internet already consists of bot activity (which is about half of all net traffic) and constant algorithmic manipulation by state and corporate actors. The prevalence of “AI slop” is evidence of this. Facebook announced that they would be deliberately flooding the site with AI characters. This bold initiative lasted all of 5 days before Zuckerberg cancelled it.

Meta removes AI character accounts after users criticize them as ‘creepy and unnecessary’

It may already be impossible — and meaningless — to wonder whether we already live in the Matrix or not. Both real and virtual worlds may both become unlivable dystopias.