At the Bloomberg Tech conference in San Francisco this week, Netflix’s chief product and technology officer Elizabeth Stone described a real consumer frustration: there’s so much content on the platform that people don’t know what to watch, when to watch it, or whether what they pick is even worth their time. The solution, according to Stone? Generative AI.
The problem she’s describing is Netflix’s own doing. Two decades of commissioning at volume created a catalogue so deep that scrolling through it became a task in itself. Now the company wants to sell you the antidote to the poison it brewed.
What they’re actually building
Stone said Netflix is already using generative AI and natural language processing to help viewers choose shows based on their mood, and is testing a voice interface alongside other experiments aimed at making recommendations sharper and more personal. The goal is an experience that’s more personalized, more interactive, and more immersive.
That last part matters more than it sounds. Netflix has long maintained that the vast majority of what subscribers actually watch comes from what the service surfaces for them — not from what they search for. Which means the recommendation layer isn’t a side feature. It’s the product. If AI can make that layer meaningfully better, it changes the whole experience of using Netflix.
The company also rolled out short clips that play in the feed and can be tapped to open a full title, save it, or share it with someone. The clip feed and the AI recommendation work share the same underlying logic: cut the distance between opening the app and actually pressing play.
The YouTube problem nobody wants to say out loud
Stone’s remarks come as YouTube continues to absorb more television viewing time — a shift that has reframed what it means to compete as a streaming service. Keeping a viewer now isn’t just about owning the title they want. It’s about being the place where they figure out what they want, fast enough that they don’t reach for a different app.
That’s the real pressure here. Netflix isn’t just trying to improve its UI. It’s trying to hold attention in an environment where the competition is essentially infinite and free.
What we don’t know yet
Stone was careful not to give a timeline. She didn’t specify which generative models are powering the recommendation experiments, and didn’t say when the voice interface moves from testing to a default feature. Netflix is in experiment mode — which could mean these features arrive next quarter or quietly disappear.
What’s clear is the direction the company is heading. The platform that made the infinite scroll a habit is now betting that AI can be the thing that finally ends it. Whether that works depends on whether Netflix is building something genuinely useful — or just wrapping the same old algorithm in a more conversational coat.