[Imagine a news program, or a whole channel, that features AI-generated anchors and reporters that look and sound exactly like real people, presenting current news stories about topics, events and people you specifically care about, from your political perspective. The stories are written by AI systems that draw on reputable sources and are accompanied by illustrations, charts and graphs, and convincing recreations of events, also all generated by AI. The story from The Hollywood Reporter below describes a startup company’s efforts to create such a channel. It’s certainly an ambitious and interesting project, but would such personalized presence experiences really be good for us and the societies we live in? –Matthew]
[Image: An AI-generated broadcast of Channel 1. Credit: Channel 1 News]
This Startup Wants to Create an AI-Generated CNN
Channel 1 News, from former ‘Tosh.0’ veteran Scott Zabielski and entrepreneur Adam Mosam, plans to use generative AI to create personalized newscasts.
By Alex Weprin
July 26, 2023
Standing in what appears to be a modern newsroom, the business correspondent breaks down the earnings report from Disney, highlighting subscriber growth at its streaming services and revenue at its theme parks. Rattling off stats and figures as images of Mickey Mouse and Disney+ fill the screen, a user might not think twice about the report.
But a closer look would show that the correspondent’s lips weren’t quite matching up to the words she was saying.
The video is a demo clip from a startup called Channel 1 News, which is seeking to use generative artificial intelligence to create a new type of video news channel. Founded by producer and director Scott Zabielski (Tosh.0, The Jim Jefferies Show) and tech entrepreneur Adam Mosam, Channel 1 will launch this year with a 30-minute weekly show made available through a FAST channel, though the ambition is to produce newscasts customized for every user (Mosam says that next year they intend to produce between 500 and 1,000 segments daily) accessible via an ad-supported app or video platform.
For now, however, the AI-generated anchors and correspondents look convincing, in the way that many AI-generated images are today, but Zabielski and Mosam acknowledge that they aren’t good enough just yet.
“The elephant in the room is this still looks like you’re watching a video game character talk. And does anybody really want to watch a video game character deliver the news to them? I don’t think so,” Zabielski says. “I think if the technology was stuck here, this would be a really hard sell. So part of what we’re doing is, like any technology, you can’t wait until it’s perfect to start on day one. So we’re obviously getting ahead of this and we’re looking down the road at 12 months from now, 18 months from now, three years from now. It is going to get to a point where you absolutely will not be able to tell the difference between watching AI and watching a human being, but we also understand that there’s going to be a pathway from here to there.”
The rationale for Channel 1, per a pitch deck for the company viewed by The Hollywood Reporter, is simple: The news, like most of the media consumed today, should be more personal. And AI is poised to help deliver that experience.
“Basically, these days, everything has sort of become personalization, whether it’s Spotify learning what you want to hear and recommending songs that maybe you didn’t know about but that you’d be interested in, or TikTok, or any of these personalized algorithms,” Zabielski says. “That’s something we don’t really see in news yet.”
And while the demo currently relies on stock footage and photos, the company intends to eventually use generative AI to re-create events that have happened, but for which cameras were not present.
“The closest analogy I could give is when you talk about a trial that was covered with there’s no cameras allowed and you’ll see the courtroom sketch,” Mosam says. “That’s kind of the best that the news has been able to do and we haven’t improved on that in decades. What we’re looking to do potentially is to add visuals where we would clearly denote this is generated imagery. So we’re not trying to pull the wool over anyone’s eyes to say like, ‘Our cameras were inside the Oval Office when this meeting happened.’”
While Channel 1 will use large language models (LLMs) to write its scripts, the company intends to source its stories from credible places. That could include press releases from verified companies (another demo sees a tech correspondent discussing Apple’s Vision Pro, with photos and videos plucked from Apple’s press site), or from partner news outlets that are interested in turning their human-generated news stories into video news reports. There will also be a team of editors verifying the accuracy of the reports.
“We’re very aware of the LLM hallucination, hallucinations and things like that,” Mosam says. “We’re staying away from that. It’s established data sources that are really trying to add a new interface on the news.”
For viewers, Channel 1 has created an AI-generated news team, with entertainment, tech, business and sports reporters.
“Imagine watching CNBC, except what you’re looking at is analysis of stocks that are in your portfolio, or industries that you’re already watching, or if you’re watching sports, it can go more in depth on the teams that you love, as opposed to waiting for the parts of the content that you’re really interested in,” Mosam says.
There are even liberal and conservative hosts who can deliver the news filtered through a more specific point of view.
“Although we can give it to you in your perspective, from your set of opinions, we will never, ever break that wall of factual reporting,” Mosam says. “So I think that if anything, we can sort of bring people together because they feel like you’re talking to them with their set of facts, opinions and demographics, but we’re holding things closer to the middle, ultimately.”
Generative AI has become a hot-button issue for the media industry, with writers and actors making it a central part of their ongoing labor strikes. The tech might not be able to produce compelling movies or TV shows on its own just yet, but no one is doubting its ability to evolve into something that could be competitive.
While Channel 1’s business models won’t be revolutionary (there will be interstitial ads, native ads, and promoted content like movie trailers), the creators hope that its content will be.
“I really do believe — I’ve been a technologist for the last 20 years, and my last startup was, 10, 11 years ago, creating a VOD platform — and I see this as the next big technological shift,” Mosam says.