Through the pixel fuzz of a sputtering Zoom connection, it’s hard to be sure if the eyes staring out from the laptop screen are human. Lars Buttler is a real person, but he is the CEO of AI Foundation, a company that makes fake ones: trainable, artificially intelligent agents that might one day take the place of a human personal assistant, a customer service representative, or you yourself (if you aspire to omnipresence or digital deathlessness). When Buttler appeared in a Zoom call last week, there was something strange about the light hitting his shaved head, and the stark white office behind him definitely wasn’t part of the material world. His speech was awkward, with overlong pauses and canned jokes that fell on a silent audience, but video calls are like that, at least for now, while people adjust to working lives forced into the ether. His eyes seemed awake and alive in a way that the faces of the other participants in the Zoom call—venture capitalist, a tech founder, and an activist, all of them puppeted by artificial intelligence—were not.

“Pretty sure Lars is human,” a (real-person) spectator typed in the in-meeting chat room.

“I’m starting to think Lars is AI,” wrote another.

The talking image of Lars Buttler, corporeal or otherwise, was the first presenter at the Virtual Beings Summit, an event celebrating advancements in creating digital entities capable of interacting with fleshfolk. The inaugural summit, in 2019, took place in a physical gallery perched just above the San Francisco Bay; due to Covid-19, this year’s proceedings were fully virtualized. It felt thematic: Gawping at Lars Buttler’s maybe-computer-generated face through a screen at home is the fairest test of what he’s supposed to represent. Conference organizer Edward Saatchi, the CEO of Fable, a company developing its own virtual being, asked presenters to provide a portal into the world they expect to see five years from now, when they think interacting with artificial people will be accepted and commonplace. Sitting in isolation, watching ambiguous faces jaw away on Zoom, that future didn’t seem as distant and dubious as it did when Saatchi said the same thing to a room full of techies last year.

It was Zoom connectivity issues that glitched away the illusion. When Digital Deepak Chopra tried to join the call, Buttler kept saying—repeatedly, in the exact same tone, volume, and cadence—“Oh! Zoom is acting weird … Oh! Zoom is acting weird.” Canny conference watchers in the chat room pronounced him virtual. “So Lars passed the Turing test for a full 15 mins. That’s got to be something,” said one attendee in the chat. “I still bet on Lars being human,” said another. Later, human Lars Buttler would pop into the call as a somewhat belated reveal, after speculation had ceased.

Virtual beings are a concept and industry so new and nebulous that no one quite agrees what they are or what they’re for. The Virtual Beings Summit applied the term to three kinds of characters: human-inhabited digital avatars (you, any time you play Animal Crossing); human-puppeteered digital avatars with little-to-no artificial intelligence (like virtual Instagram influencer Lil Miquela, who is a static image most of the time); and more autonomous-appearing, artificially intelligent digital beings (like the phony Lars Buttler).

Without the tempering influence of meatspace interactions, and with participants uniquely free to chat during presentations, people weren’t shy about their confusion. Many wondered about what they were being shown, where it fit in the taxonomy of virtual beings, whether they were watching something live or prerecorded, whether a human was conducting the performance somewhere or if the artificial person could really speak for itself. “We need to distinguish between the level of polish of the presentation, which is obviously very high, and what’s ‘under the hood’ in terms of the AI,” an attendee said. Distinguish we did not. Companies behind fake people don’t like to get real.