Here's a fun thought experiment: What happens when AI models trained on internet content start encountering content created by other AI models? According to Dan Houser, co-founder of Rockstar Games, you get the digital equivalent of mad cow disease.

Speaking on Virgin Radio UK's "The Chris Evans Breakfast Show" on Nov. 26, Houser laid out his concerns about artificial intelligence's self-cannibalizing future. "The internet is going to get more and more full of information made by the models, so it's kind of like when we fed cows with cows and got mad cow disease," he explained. "I think that AI is gonna eventually eat itself."

It's not just the technology that worries him. The people championing widespread AI adoption aren't exactly inspiring confidence either. Company executives pushing AI forward are "not fully rounded humans," Houser said bluntly.

"Some of these people trying to define the future of humanity, creativity, or whatever it is using AI, are not the most humane or creative people," he added. "So they're sort of saying, 'We're better at being human than you are.' It's obviously not true."

Fiction Meets Reality

Houser appeared on the show to discuss "A Better Paradise," his debut novel. The book features an AI model that becomes "more sentient" than its developers wanted while simultaneously failing to meet business expectations. Sound familiar?

Even though the novel is set in the 2030s, Houser said real-world events have been catching up to his fiction faster than expected. "I would've never thought I was that good at predicting the future," he noted. "If anything, it should be set nearer to today."

The Confidence Problem

While Houser acknowledges AI has useful applications, he's bothered by some fundamental issues with the technology. The inconsistency particularly gets under his skin.

"I'm slightly obsessed with the fact that when you search the same thing again it doesn't give you the same answer," he said. "And it's wrong a lot of the time but it says it so confidently."

That combination of unreliability and unwavering confidence might just be the most human thing about artificial intelligence after all.