At the end of September, YouTube announced a major update that could reshape how videos are discovered, consumed, and promoted on the platform. Users will soon be able to craft their own recommendation algorithms: simply describe what you want to watch, and AI (powered by Gemini) will deliver a unique, personalized feed. Here’s what this means for AI creators, marketers, and anyone working with video.
How YouTube’s New AI Recommendations Work
Traditional Recommendations vs. Personalized Feeds
Previously, YouTube suggested videos based on your general activity—what you watched, liked, or subscribed to. While this caught some of your interests, the one-size-fits-all algorithms couldn’t always predict exactly what you wanted. The new feature changes everything: just type a request like, “Show me the latest videos about generative AI and business use cases,” and Gemini-powered AI will assemble a tailored stream for you.
Usage Example
Instead of endlessly scrolling the default feed, a user can type:
“Show me neural network videos for SMM, only from 2024, no long lectures—maximum 10 minutes each.”
In seconds, AI curates a playlist that’s laser-focused on these preferences.
Why YouTube Is Launching This Now
Competition for online attention is fiercer than ever. TikTok, Instagram, Shorts, Reels—all are fighting to keep users. Hyper-personalization is the logical next step. By integrating Gemini, YouTube aims not just to retain viewers, but to make finding the right content effortless.
Opportunities for AI Creators and Marketers
New Promotion Scenarios
- Target by Query. Your job isn't just to land in trending lists anymore; it’s to anticipate the specific questions and phrases your audience will use. The more accurately you match their queries, the higher your chances of appearing in their personalized feeds.
- Long Tail Visibility. Niche videos that once got buried now have a shot at discovery. If someone’s looking for something specific, your specialized content can finally reach them. This is a huge opportunity for creators in narrow verticals.
- Format Experiments. Try different styles and titles to match varied user prompts—from “Explain it like I’m a beginner” to “Give me a professional analysis.”
A Shift in Optimization Strategy
Tags and descriptions aren’t enough anymore. You need to understand how people phrase their requests in natural language. That means optimizing for AI-driven recommendations now involves semantic analysis, keeping up with trending queries, and constantly collecting feedback.
New Challenges for Marketers
- Classic A/B tests for thumbnails and titles become less important—real user queries become your new testing ground.
- There’s a new KPI: how often does your video appear in personalized feeds for relevant queries?
- Understanding your audience’s language—how they describe their needs and interests—becomes essential.
What to do right now
- Start analyzing user queries. Use trend tools and monitor how your viewers phrase questions in search and comments.
- Review and diversify your video tags and descriptions. Make them relevant not just for standard keywords, but for “human” language and natural phrasing.
- Experiment with formats. Release explainer videos, reviews, guides, and case studies—and see which ones land in the new AI-driven feeds.
- Leverage AI analytics. Try insight generators for YouTube data to quickly spot trending topics and query phrases.
- Keep up with platform updates. YouTube will keep evolving its AI personalization. Be among the first to adopt new features.
More hands-on AI tools and templates for video—find them in my courses.
Based on reporting by TechCrunch


