If the future of the internet looks like a constant stream of funny videos produced by artificial intelligence, Openai has just placed its stock in emerging markets.
The company behind ChatGPT released a new SORA social media app on Tuesday. This was trying to attract attention from eyeballs currently staring at short form videos on Tiktok, YouTube, or meta-owned Instagram and Facebook.
The new iPhone app takes advantage of the charm of being able to make videos of yourself doing everything you can imagine from anime in a very realistic style.
But the video scrolling flood that takes over social media is worried about “AI slops” that flock with more authentic human creativity and degrade the information ecosystem.
“These things are very persuasive,” said Jose Marichar, a professor of political science at the California Lutheran University, who studies how AI is being reconstructed. “I think sucking you means they’re kind of incredible, but they look realistic.”
The official release video for the SORA app features an AI-generated version of Salewing Openai CEO Sam Altman from the Psychedelic Forest. The Moon and Stadium are then crowded with cheerful fans watching rubber duck races. He introduces new tools before handing over to a colleague placed in other quirky scenarios. The app is only available on Apple devices for the time being, starting in the US and Canada.
Meta launched its own feed of AI short form videos within the Meta AI app last week. In an Instagram post unveiling the new Vibes product, Meta CEO Mark Zuckerberg posted a carousel for AI videos. Both Sora and Vibe are designed to be highly personalized, and recommend new videos based on what people are already involved in.
Marichal’s own social media is fed on Tiktok and other sites. It’s full of videos like that, from “house cats riding wildlife in terms of doorbell cameras” to fake natural disaster reports that are both fascinating yet easily exposed. He said he can’t blame people for trying hard, “I want to know if something extraordinary is happening in the world.”
The danger, he said, is when they control what we see online.
“We need an information environment that is almost true or reliable, because we need to use it to make reasonable decisions about how to govern collectively,” he said.
If not, “we’ll be very skeptical or very certain about everything,” Mallical said. “We are either manipulated or manipulators. It leads us to something that is non-specially liberal democracy.”
Openai made some efforts to address these concerns in its announcement Tuesday.
“Concerns about Doomscrolling, Addiction, Isolation and (Renforcement Learning) slop-timed feed are at the top,” he said in a blog post. It said it would “vote users regularly about their well-being” and provide them with the option to adjust their feed, with a built-in bias to encourage posts from friends rather than strangers.
————
AP Technology Writer Barbara Ortutay contributed to this report.
