What Spotify’s New AI DJ Can Teach Video Streaming Services

Radio "On Air" sign
Illustration: VIP+; Adobe Stock

I don’t know Xavier “X” Jernigan, head of cultural partnerships at Spotify, but he certainly seems to know me. 

Jernigan has been in my ear a lot lately as I’ve listened to Spotify via a new feature his employer deployed in beta on the app last week. He has been talking to me between the songs he recommends I listen to based on his extensive knowledge of my musical tastes.

Only, Jernigan isn’t actually talking to me at all. He has lent his voice to Spotify’s AI DJ, which puts a fresh spin on the company’s algorithm-driven recommendation technology by expressing it in the form of a human-voice simulation that is customized to each user.

In terms of its algorithmic prowess, AI DJ is probably no more effective at serving up personalized song recommendations than anything Spotify is already currently doing on its platform. Regardless, what makes AI DJ stand out as an innovation is its vocal manifestation. There’s something uniquely powerful about taking the same old algorithms and anthropomorphizing them in a human-sounding voice. 

You won’t hear that halting, stilted speech you’re accustomed to from Siri or Waze. It’s a remarkably authentic voice created by Spotify acquisition Sonantic, the tech firm responsible for restoring Val Kilmer’s voice in “Top Gun: Maverick.”

And while Jernigan’s smooth patter may indeed sound like that of your local-station DJ, this isn’t some conglomerate pumping the same message to millions of listeners regardless of their tastes. He speaks to you in a way that constantly reflects he knows what music you like by name-checking your favorite genres, bands and songs at every turn. 

Spotify needs all the help it can get boosting subscriber engagement given the stiff competition it gets from rival music services. But maybe there’s a valuable lesson for its counterparts in the premium-video business, too. 

Imagine a world in which the homepage of your go-to video streaming service wasn’t eye-glazing stacks of tiles promoting individual TV and movie titles separated into themed collections that don’t seem to have any connection to your individual tastes.

Imagine, instead, a homepage experience featuring an AI personality with intimate knowledge of your viewing habits. This figure greets you by name, wishes you a “good afternoon” because it knows you’re watching at 2 p.m. and references the last program you watched as it asks if you’d like to pick up where you left off.

And if you were in the market for something new to view, maybe an AI-fueled avatar is up to the challenge. “Hey, I know you liked ‘Breaking Bad’ because you binged those episodes faster than any show you’ve ever watched before,” this imaginary video DJ tells you. “Are you ready to try the prequel to the show, ‘Better Call Saul?’ Or maybe, because you usually watch comedy in the afternoons, I can interest you in Bryan Cranston’s lighter work ‘Malcolm in the Middle’?”

Why do streaming video services need to think about something like AI DJ? Consider what Variety Intelligence Platform explored last week with regard to what we think is streaming’s biggest struggle: content discovery. There is just way too much UX friction on these platforms, and it’s going to take a bold solution to address the problem. 

For all the hype Netflix generated about its algorithm-driven recommendations, the truth of the matter is it isn’t all that impressive. Yes, Netflix is clearly better than its competitors at personalizing the user experience, but a much better version is possible. Just think about how much better Spotify or TikTok is at sucking users in by studying their habits and serving up a content mix that keeps them glued to their apps.

In a competitive set in which more than a handful of streaming services are spending billions on content to beat one another’s brains out, is it so crazy to think content doesn’t necessarily have to be the differentiating factor that allows a market entrant to stand head and shoulders above the pack? 

Think about your favorite bar or pub. It’s not the alcohol that keeps you coming back; it’s the ambience of the establishment in which it’s served and/or the bartender who serves it up with a knowing wink.

Of course, curating audio and long-form video isn’t the same thing, but that shouldn’t keep streaming services from trying to glean lessons from the likes of AI DJ. And short-form video platforms like YouTube or TikTok may have an even higher upside to explore. The same goes for FAST platforms such as Pluto TV; linear-channel presentations are back in vogue, but that doesn’t necessarily mean they are ideal, either. 

There’s this presumption among digital platforms of all kinds that algorithmic recommendations are best experienced operating silently and invisibly in the background, but what if the reverse is true? What if the stickier way to go is to shift content consumption from a purely solitary experience to one that pairs the user with a virtual friend intent on finding him or her the right piece of entertainment? 

I’m not saying that’s what AI DJ is for Spotify, at least not in its current form. AI DJ is not all that sophisticated, just sophisticated enough to get you thinking about how much more impressive it could get as the underlying technology improves. 

For instance, you don’t actually see Jernigan in AI DJ—he's just a green circle on a blue background that pulses as he speaks. But I wouldn’t be surprised if Spotify eventually figures out how to visually re-create him as well, given how advanced so-called deep-fake technology is today. 

And surely the time will come when Spotify will empower AI DJ to not only do the speaking but be spoken to so there’s two-way communication. It’s a reminder that generative AI isn’t the only revolutionary technology poised to sweep through the entertainment world (and every other industry out there). Voice-command tech will be a revolution unto itself. Together, they’re a game-changing combination. 

Perhaps one day not so far in the distant future, one streaming service will distinguish itself by offering a user experience so superior to what’s available today it’s the attribute that propels its success more than anything else. It’s hard to conceive, but it’s not inconceivable.