YouTube is experimenting with an AI chatbot aimed at giving users more control over the chaotic Home feed. Some users are seeing a new “Your Custom Feed” chip next to the Home tab, letting them tell YouTube directly what kind of videos they want more — or less — of. The feature hands users the steering wheel, replacing passive algorithm tweaks with an interactive feedback loop.
How the Chatbot Shapes Your Home Feed
Early testers can tap the “Your Custom Feed” chip and enter prompts like “More long-form explainers about astrophysics” or “Fewer prank videos and celebrity gossip.” The chatbot interprets these instructions and instantly reshapes your Home feed around your stated interests — no need to dig through settings or guess how to “train” the system through likes or clicks.
While YouTube hasn’t detailed the entire process, the concept relies on its existing recommendation model, which uses deep neural networks to match user interests with video attributes. Adding a conversational layer allows the system to refine these “interest embeddings” more precisely than past feedback tools such as watch time or “Not interested” taps.
Importantly, the chatbot doesn’t reset your feed entirely. It’s more like a rapid tuning option — type “more cooking tutorials, fewer Shorts, no crypto,” and the changes appear almost immediately. This could help stop users from spiraling down rabbit holes where a single curiosity click skews recommendations for weeks.
Why YouTube Is Reworking Its Recommendations
With over two billion monthly logged-in users, YouTube’s Home feed is one of the most influential spaces online — and frequently a point of frustration. A stray click can lead to an endless stream of loosely related videos that’s hard to shake off.
Research from Pew and the Mozilla Foundation shows that recommendation systems often shape what people see far more than search does, and that algorithmic suggestions sometimes lead to irrelevant or problematic content. YouTube has tried to fix this with feedback tools like “Don’t recommend channel” and topic filters. The new chatbot goes further, turning vague preferences into explicit instructions that the algorithm can act on instantly.
This move also follows a broader industry trend toward transparency and user empowerment in recommendation systems. Under growing regulatory scrutiny, platforms are finding new ways to show that personalization doesn’t have to mean opacity.
How It Stacks Up to Other Platforms
Other platforms have tested AI-driven personalization with mixed success. X introduced an AI-curated timeline via its Grok model, Spotify built an AI DJ that talks users through selections, and Netflix enhanced its thumbs rating with a “Double Thumbs Up.”
YouTube’s approach, however, stands out for its immediacy and scale. The Home feed is already the platform’s main discovery engine for both viewers and creators. A responsive, conversational editor at that entry point could make personalization far more intuitive. Instead of spamming “Not interested” buttons and hoping the algorithm catches on, you can simply say, “Show me more beginner-friendly Python tutorials” or “Tone down political commentary,” and watch your feed adapt on the spot.
What Testers Should Keep in Mind
As with any experiment, expect limitations. The feature appears in support documents but is rolling out to a small group, and YouTube may adjust or drop it depending on results. If you do get access, be specific in your prompts: mention topics, creators, video lengths, or styles, like “More 20–30-minute climate tech documentaries this week; fewer Shorts and reaction videos.”
Your watch history will still matter. If it’s paused or shared with others, steering accuracy may suffer. Broad prompts can also cause abrupt shifts, so it’s best to combine granular inputs with existing feedback options such as “Not interested,” “Don’t recommend channel,” and curated subscriptions.
Privacy remains an open question. Since conversational prompts can influence your profile, users may want clearer information on how those inputs are stored or used. Transparency around data retention — especially for younger or vulnerable users — would strengthen trust.
The Bottom Line
If YouTube can successfully scale this test, the Home tab could evolve from a mysterious algorithm into a customizable tool. For creators, it may mean being discovered through audience intent rather than algorithmic luck. For viewers, it promises an escape from the recommendation spiral — one conversational prompt at a time.
