Chris Hayes Says AI Finds Viral Hits ‘No Executive’ Ever Would, Algorithms Are ‘Learning at Scale’

The AI deployed by social media platforms is “learning at scale,” MSNBC anchor and author of “The Siren’s Call: How Attention Became the World’s Most Endangered Resource” Chris Hayes told Slate’s “What Next” podcast in an interview published Sunday.

AI is given data and learning off it “independently,” he added, and can churn out content that “no executive would have ever greenlit” but that appeals to the public in a way that corporate executives “never have to figure out.”

Hayes and host Lizzie O’Leary were focused on infinite scroll, the mechanism by which so many smartphone users use the internet and social media. The infinite scroll “means you never have to stop,” Hayes explained. “I also posit that it feels like a slot machine for nonaccidental reasons. The slot machine is probably the single most successful pure attention-monetizing technology ever derived. People can spend eight hours there with these little five-second bursts.”

The infinite scroll in turn feeds the machine, which has an apparently limitless capacity for information. The algorithm, he continued, is where AI as we currently understand it was “first deployed truly at scale” in a “customer-facing way.” In other words, he added, “what the social media algorithm is doing is just machine learning at scale.”

The AI takes in “a lot of data” and learns “off that data independently.” One area AI is exposed to is the attention we each choose to give social media and the content we consume. “I mean, there’s all sorts of weird genres, horse hooves being cleaned out, pipes being unclogged, carpets being cleaned that turn out to get people’s attention quite reliably that the algorithm just figured out,” he said. “But no executive would have ever greenlit [that content]. And this is actually a really big difference in this technology. They never have to figure out. Netflix, even with its algorithm, still has executives saying yay or nay.”

But executives at social media companies can just trust the machine they’ve made to do the work for them. For example, the team at TikTok will “never have to figure out what’s going to work or not. Everything just gets thrown against the wall and the machine learning figures it out.”

This isn’t inherently an unpleasant thing, Hayes also said. “There’s cool aspects of that because you discover stuff that you wouldn’t maybe have otherwise thought would work.”

But the social component, such as “the hard wording of the mentions, the fact that notifications are weaponized” is concerning. The best example of this, he said is “the haptic feedback of a phone, a buzz in the phone, is the ultimate example of involuntary attention.”

“You will feel that go off in your pocket. You don’t really get to say if you do or do not. You engineer all that together, and you have the most sophisticated machine for keeping people’s attention that ever existed.”

Hayes acknowledged that as a broadcast journalist, he participates in a similar attention-grabbing practice to some degree, but that “the attention is a means towards some end.”

“I am not just on air to keep people’s attention,” he continued. “That’s not the project I’m engaged in. And if I were, I might make different choices. What is different about the machine learning algorithm over at Bytedance or Instagram is that it does purely do it for the purpose of attention. There is no other purpose. It’s driving towards nothing.”

You can listen to the entire interview at Slate.

The post Chris Hayes Says AI Finds Viral Hits ‘No Executive’ Ever Would, Algorithms Are ‘Learning at Scale’ appeared first on TheWrap.