Google continues to push artificial intelligence into every possible corner of its ecosystem, and the next target is YouTube. As part of its new experiment, the company is launching a feature called "Ask YouTube" – a search tool that aims to change the way users find content on the platform. Instead of settling for a list of videos, users can ask complex questions and receive combined answers of text, video, and guidance for further exploration.

The feature is currently available in a limited rollout to YouTube Premium subscribers in the United States, over the age of 18, and will operate until June 8 as part of the YouTube Labs testing program. To use it, it must be enabled through the settings, after which a new button will appear in the search bar. From there, users can choose ready-made suggestions or type free-form queries – for example, planning a three-day trip between San Francisco and Santa Barbara.

The result, at least according to initial tests, is an experience that tries to feel like a conversation: A summarized answer, relevant video clips with timestamps, and the option to ask follow-up questions. For example, a search about the history of the Apollo 11 landing presented a summary alongside focused videos. However, not every query yields "smart" answers – sometimes the result is simply a regular list of videos, as would be received in a classic search.

As expected, the familiar problems of AI also arise here. In one case, according to a test by The Verge website, the system provided completely incorrect information. This is a reminder that despite the progress, the technology is still not fully stable.

In the background, the gap between the enthusiasm of tech companies and the sentiment of users is particularly noticeable. The YouTube audience is known to be suspicious of content that is created or mediated by artificial intelligence, and sometimes even hostile toward it. Therefore, the success of the new feature will not be measured only by its technical capabilities – but by a simpler question: Does it really help find good content faster, or does it just add an unnecessary layer on top of what already works.