Netflix Uses Generative AI to Overcome Content Overload It Created

At the Bloomberg Tech Conference

At the Bloomberg Tech conference in San Francisco, Elizabeth Stone, chief product and technology officer at Netflix, stated that the company is using generative AI to help subscribers navigate the vast amount of content accumulated over twenty years of catalog. The problem encountered, in her words, is "a mounting frustration: there is so much content. How do I make sense of it, and what is right for me, and what is right for me at this moment?"

The direction is not new. Last year, Netflix had already introduced a revamp of the interface with a generative AI-based conversational search, then in beta on iOS, capable of responding to natural language requests like "something fun and light." The statements at Bloomberg Tech represent the desire to extend that idea.

The Discovery is the Product

Stone explained that generative AI and natural language processing are already used to suggest titles based on the viewer's mood, and that the company is experimenting with a voice interface along with other tests designed to refine suggestions. The goal is a "more personalized, more interactive, more immersive" experience.

Netflix has always claimed that the vast majority of what subscribers watch comes from what the platform recommends, not from what they search for. This makes the discovery moment the part of the service most exposed to a good idea from others.

Pressure from YouTube

And that good idea could come from a competitor. Stone's words come as YouTube continues to absorb television viewing time, a shift that has redefined the demand for every subscription service. Retaining a viewer no longer depends solely on possessing the title they want: it matters to be the place where they decide what to watch, quickly enough not to open another app.

The push for discovery accompanies a series of interface modifications that Netflix has been testing for months, including short clips scrolling in a feed that can be tapped to open a title, save it to a list, or forward it to someone. The clip feed and the work on suggestions share the same logic: shortening the distance between opening the app and hitting the play button.

There is irony in the strategy: the choice paralysis that AI is supposed to solve largely arises from the company's own catalog policy, from years of high-volume productions that filled the grid faster than any person could organize it.

Stone did not indicate a timeline for a broader release of the voice interface, nor which generative models are behind the suggestions. Netflix claims to be experimenting, but has not clarified when the experiments will become defaults.