More AI: Will YouTube’s new promptable feed replace the recommendation algorithm?

YouTube and X may be closer than they first appear: Google‘s video platform is testing a new type of feed where users will be able to use text prompts to shape their personal recommendation algorithm.

This sounds just like the promises made in September by Elon Musk. People–content creators and viewers alike–have complained about recommendation algorithms for as long as recommendation algorithms have existed, but after Musk bought X, those complaints had some extra justification. In the years following his purchase, X’s main feed algorithm deteriorated into slop, and on top of that has been (allegedly) tweaked to push posts from Musk and amplify right-wing views. Many users report a deluge of hate speech and violent videos showing gruesome fights and animal deaths, too.

Musk’s plan to combat user concerns is bringing in Grok, the ChatGPT-alike made by his company xAI. According to Musk, X’s recommendation algorithm will soon be “purely AI,” and “[b]y November or certainly December, you will be able to adjust your feed dynamically just by asking Grok.”

That hasn’t happened yet–at least, not on X.

But over on YouTube, only a week after Google CEO Sundar Pichai warned that the amount of money being poured into the AI bubble is “irrationality,”

an experimental feature will allow users to “update your existing Home feed recommendations by entering a simple prompt,” Team YouTube wrote on the platform’s support/developer update notice board.

This also comes just after Google rolled out the third iteration of its own ChatGPT equivalent, Veo. We’re not sure if this custom feed feature will be built on Veo’s bones or even have Veo directly involved, the way X has Grok–but either way, Google’s investment in AI development is being shown in tangible form here.

“If you are part of the experiment, you will see ‘Your Custom Feed’ appear on your Home page as a chip besides ‘Home,'” Team YouTube wrote. “This feature is designed to give you an easy-to-use way to have more control over your suggested content. If you see it, check it out and share your feedback!”

As we mentioned above, the YouTube algorithm has long been a point of contention. Creators often have an antagonistic relationship with it, and have accused it of shadowbanning them and refusing to recommend their videos to subscribers. The algorithm has also been accused of recommending disturbing videos, sending users into conspiracy rabbit holes, and feeding viewers’ mental health and body image struggles. There are plenty of studies about how social media recommendation algorithms–not just YouTube’s–can negatively affect viewers.

But, at the same time, YouTube’s algorithm has unique charm. Anecdotally, I’ll often open my Home page and see 20-year-old flip phone recordings suddenly revitalized by an inexplicable algorithm boost. Other times, my partner and I will bond over being recommended the same video on the same day because we have niche intersecting interests (the operational minutiae of offshore oil platforms, anyone?), though the rest of our feeds’ components are pretty different. The algorithm has also, of course, introduced me to new creators, and at the same time resurfaced artists I used to know on DeviantArt.

There is a certain magic to the recommendation algorithm being undisturbed in its natural habitat. But Google, like every other big tech company these days, wants the AI bag bad, and we can see how this might be a perk for some viewers (especially those who mostly watch Shorts; we’ll admit that algorithm is, objectively, garbage).

Like any YouTube experiment, this new feed might not be rolled out as a real product to all users. Even if it doesn’t, its testing phase will give Google data about how people use chatbots to shape their online experiences…for better or worse.