Facebook will let you customize your feed algorithm using AI
Meta is introducing Your Algorithm on Facebook, allowing users to shape the topics recommended in their feed and Reels. The feature matters because it gives people a direct way to tell Facebook what they want to see, with recommendations that Meta says adjust in real time.
Users can open the feature or mark individual posts as “interested” or “not interested”, including through a post’s three dots menu. Meta may ask follow-up questions and let people describe in their own words what they want more or less of. The tool follows similar features on Instagram and Threads, and Meta says it is using large language models in its recommendation systems.
- Facebook users can tune recommendations by topic.
- Meta says changes update recommendations in real time.
- The feature builds on similar tools on Instagram and Threads.
New here? Start with this
Facebook is a social network owned by Meta, which also runs Instagram and Threads. Its feed and Reels use recommendation systems to choose which posts and short videos to show each person, based partly on their activity and interests.
Meta’s recommendation systems have traditionally shaped what people see without users directly setting every preference. Similar controls on Instagram and Threads let people indicate topics they want to see more or less of, and Meta says it uses AI language models to help tailor recommendations. These tools matter because recommendations influence how users discover posts and spend time on the platforms.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Giving users direct control over recommendations could make Facebook more relevant to their interests and reduce unwanted content. Natural-language requests and quick feedback may make personalisation easier to use, while real-time adjustments could give people more agency over their feeds.
The case against
A feed shaped by users’ stated preferences could become narrower, showing less variety and fewer perspectives than a system that balances personal interests with broader discovery. The feature also relies on Meta’s recommendation systems and its interpretation of users’ requests, so people may have limited insight into how their choices affect what they see.