YouTube tried replacing creators with AI slideshows, viewers hated it, so it built something smarter

YouTube ran an experiment last summer in which AI selected segments from creator videos, stitched them into slideshows, and served them as answers to search queries. The feature launched to US Premium subscribers in June 2025 and expanded in August. Then it quietly vanished.

“People just didn’t like it,” Emily Moxley, YouTube’s vice president of product management, told Semafor. Viewers wanted to hear from the creators themselves, not consume their knowledge stripped of context and personality. The experiment proved that the relationship between viewers and creators is not something AI can easily replace.

YouTube went back to the drawing board and recently launched Ask YouTube, a conversational AI search feature that enhances discovery without pushing creators into the background. The tool was announced at Google I/O in May for Premium subscribers, expanded to US desktop users this month, and is coming to mobile and TV soon.

Under the hood, Ask YouTube breaks a complex question into simpler ones, fans them out across YouTube’s existing search systems, and organises the results. Rather than generating a new video from stitched clips, the feature jumps users straight to the specific moment in a creator’s video that answers their question. The AI writes only the connective text between results.

Sandie Gong, the product lead for Ask YouTube, said the feature is changing how people search. “We’re seeing a lot more of these complex queries where people are clearly sharing the full extent of their underlying intent,” she told Semafor. Users are typing detailed questions like how to fix a specific model of garage door opener, or asking to see the view from a particular seat inside the Las Vegas Sphere.

The idea emerged from YouTube’s existing learning hub, where the team noticed users were ignoring the tool’s intended educational purpose. People were typing queries like “plan me a seven-day trip to San Francisco” and “how to survive a zombie apocalypse,” revealing demand for a conversational search experience that YouTube’s traditional keyword box could not satisfy.

The stakes for getting this right are enormous. YouTube is now the single biggest distributor of television time in the United States, according to Nielsen. MoffettNathanson estimates the platform took in roughly $62 billion in revenue last year, more than any other media company in the world.

Its most recent quarter delivered ad revenue of just over $11 billion, up 13 percent year on year. The platform is sitting on 20 billion uploaded videos, almost all of them made by human creators whom YouTube has paid more than $100 billion over the past four years. That library is a potential goldmine for AI training, particularly for robotics and autonomous vehicles, but exploiting it too aggressively risks alienating the people who built it.

YouTube’s own data supports the bet on human content. The failed slideshow experiment showed that even using AI to edit and recombine creator footage was a step too far for viewers. A cottage industry of “Ask YouTube SEO” guides has already sprung up among creators looking to optimise for the new search format, and Ask YouTube views count toward monetisation.

Moxley said AI-generated videos receive no special treatment in Ask YouTube results. “We want to give people the videos that they want to watch,” Gong said, “however it’s created.” But the platform has simultaneously been cracking down on AI-generated content at scale, terminating channels and automatically labelling synthetic videos.

The tension YouTube faces mirrors a challenge across the entire tech industry. Adopting AI wholesale risks disrupting a business that already works, but moving too slowly risks disruption from competitors building AI-native alternatives. YouTube’s answer, for now, is to use AI as a lens that makes human content easier to find rather than as a replacement for it.