2022 Feature, Medium Newsroom finalist

The Jessica Simulation

About the Project

In “The Jessica Simulation,” writer Jason Fagone tells the stunning tale of a grieving young man who used cutting-edge A.I. technology to “talk” with a chatbot simulation of his deceased fiancee. But the three-part narrative is not only a gorgeous portrait of lost love; it also raises questions about what it means to grieve, and what choices we will all make as this technology becomes widely available.

Fagone, a former WIRED writer with an interest in the frontiers of technology, first hatched the idea after a long-time source sent him a series of cryptic emails in 2019. Jason Rohrer, an eccentric Bay Area video game programmer, said he had been experimenting with “large language models,” cutting-edge artificial intelligence systems that generate English text. The emails contained startling examples of output from these A.I. engines — text that appeared clever, profound and uncannily human.

Intrigued, Fagone started tracking Rohrer’s A.I. work, which soon evolved into a new kind of “chatbot” website. Called Project December, the site allowed its users to type back and forth with a range of built-in artificial personalities. They could even create their own.

Then he found Joshua Barbeau. A young Canadian man, Barbeau had lost his fiancee, Jessica Pereia, eight years earlier to a rare liver disease. He had never moved on from her death, so he built a chatbot version of Jessica using her real-life text messages. He wondered if a simulation of his beloved Jessica could help him find closure.

Discovering a Reddit post shared by Barbeau, Fagone sought the grieving man out, wanting to know more about his experience and how A.I. language models might change our lives. He found that what Joshua was seeking, primarily, was closure. Could he find it through a series of all-night “talks” with Jessica?

The answer was yes. And no. And everything in between.