The startup is developing a news service based on artificial intelligence. Will this change a decade of parasocial relationships between TV audiences and the people they watch on screen? asks the BBC

For 22 minutes, several news anchors sit in front of the camera and review the news of the day in the published video in social networks. But none of them are real people, they are created by artificial intelligence (AI).

The video was produced by Los Angeles-based Channel 1, a startup founded by entrepreneurs Adam Mossam and Scott Zabielski, which will launch AI-powered news on a streaming TV channel later this year. “This is a very exciting opportunity to improve the news experience,” says Mossam.

AI technology can also help translate scripts and interviews from one language to another, a capability that Channel 1 demonstrated in a promotional video released in December.

Channel 1 is the latest showcase of AI news anchors from around the world. In Kuwait, an AI-created person named Fedha made headlines in Kuwait News. Hermes presented the news in May 2023 for the Greek state broadcaster ERT. South Korean broadcaster SBS has outsourced its news coverage to Zae-In, a deepfake created by AI. There are others in India and Taiwan, all created by AI.

But one key question remains to be answered: Will viewers trust news delivered by AI over humans?

“We’re not doing this because we think a robot can do a better job than a human,” says Adam Mossam.

According to an Ipsos survey, trust in people reporting the news has fallen to an all-time low. Only 42% of people in the UK trust TV newsreaders, down 16 percentage points on the year. Skepticism about news anchors as independent arbiters of truth is a rare modern phenomenon, and many choose to get their news from content creators or influencers.

These social media stars use a connection with their audience called parasocial influence. First discovered in the 1950s by scientists at the University of Chicago, parasocial relationships were defined as the belief that viewers of the nightly news broadcasts of the time believed that an on-screen TV star was speaking directly to them through a camera. News anchors have become more than just journalists; they were friends welcomed into the living room at home.

Social media influencers adopted the same format and made a personal connection with great success. “It’s interesting how the label ‘parasocial’ has evolved from describing the kinship people can feel with news anchors at a distance,” says Christine Tran, who studies digital platforms at the University of Toronto. “You can have a parasocial relationship with a reporter or a Twitch news streamer,” she says, referring to the live video streaming service. “But you can also—probably—have a parasocial relationship with a YouTuber, a singer, and an Instagram couple.”

But whether artificial intelligence can recreate a personal connection is less certain. “You’ll never have the same connection with an AI as you would with another human,” Mossam admits. However, he argues that people are no longer looking for impartiality. “We’re not doing this because we think a robot can do a better job than a human—that’s ridiculous.”

“Will AI personalities create the same parasociality if their shows are labeled as ‘AI content’ and viewers know there’s no personal life behind the scenes?” researchers ask.

Currently, Channel 1 has about 12 employees who review AI-generated scripts and choose which stories to cover. Mosan says Channel 1 goes through a 13-step process for each story before it airs to make sure there are no problems with the generative AI. The company behind it is looking to hire an editor-in-chief early next year.

Mosan and Newman agree that the ability to find newsworthy events is another element that AI can struggle with. Channel 1’s test episode relied heavily on stories and footage shot by human journalists. “Without those sources, I don’t really see how I could do it,” Newman says. “If this raw material is not available, artificial intelligence has nothing to work on.”

Mosan believes there are some elements of the reporting process that can be done by artificial intelligence, but others cannot. “You’re never going to be able to gather information from person to person and interview the person effectively,” he says. “But I could fly a drone and analyze what I’m looking at.”