The Curious Case Of Elias Thorne

Elias Thorne has lived many lives. He has been a lighthouse keeper, a clockmaker and a detective. Depending on which AI-generated story you read, his profession, location and circumstances may change. Yet one thing remains remarkably consistent. He keeps appearing.

The strange part is that Elias Thorne doesn’t exist.

Researchers recently analysed large numbers of AI-generated stories and found that more than a quarter featured the name Elias. Many also reused the same small collection of locations, occupations and narrative patterns. Rather than producing an endless variety of fictional worlds, the models often returned to familiar characters and settings.

At first glance, it feels like little more than an amusing quirk. But it also hints at something more significant. As AI-generated content becomes more common online, future AI models may increasingly learn from material produced by earlier AI systems rather than entirely human-created sources.

The concern isn’t that AI invents fictional people. Stories have always been full of memorable characters. The concern is what happens when AI starts learning from its own habits. If the same names, ideas and story structures appear over and over again, the range of outputs may gradually become less varied.

Researchers sometimes describe this as model collapse: a process in which AI systems trained on increasing amounts of synthetic data begin to lose diversity over time. Instead of generating fresh combinations, they become more likely to repeat familiar patterns because those patterns already dominate the information they’ve learned from.

Elias Thorne isn’t the problem. He’s simply a memorable illustration of what can happen when AI begins echoing itself.

It’s a useful reminder that the future of AI depends not only on how much information models are trained on, but also on the variety of that information. Diversity of data doesn’t just influence accuracy. It also influences originality. Elias Thorne may never have existed. But the fact that AI keeps inventing him tells us something very real about the future of artificial intelligence.