Netflix Rolls Out New AI Feature Powered By OpenAI

Netflix has begun rolling out a new AI-powered feature designed to help users discover movies and TV shows more easily. Developed in collaboration with OpenAI, the feature is currently available for testing to select subscribers in Australia and New Zealand. It is expected to launch in the United States and other countries soon.

According to a Bloomberg exclusive, the new AI feature allows users to search for video content considering multiple new factors, including personal elements such as mood. Considering these inputs, the AI will suggest options available in the user’s catalog.

Users in New Zealand and Australia can test the new feature on IOS devices, and Netflix is working on expanding to more regions, including the United States. Those interested in testing the AI tool must opt in once it becomes available in their area.

While Netflix already uses AI for its algorithm to consider subscribers’ history and liked shows and movies for new suggestions, the new AI feature is expected to go a bit further by allowing users to pose more complex inquiries and use natural language.

The entertainment company also mentioned that they are using AI technologies in the production of movies and shows, but clarified that they are not replacing screenwriters, actors, or other creative workers. The use of AI in creative environments has been a delicate topic in the past few years, and a source for debate and new agreements between tech companies and organizations such as the Hollywood union SAG-AFTRA.

A few days ago, Netflix’s engineering team also shared details of its new internal research to update the platform’s personalized recommendation system on Medium. The team’s Foundation Model for Personalized Recommendation considers Large Language Models (LLMs) and a specialized tokenization system to enhance its suggestions and optimize processes.

“The Foundation Model allows various downstream applications, from direct use as a predictive model to generate user and entity embeddings for other applications, and can be fine-tuned for specific canvases,” states the document. “This move from multiple specialized models to a more comprehensive system marks an exciting development in the field of personalized recommendation systems.”