At a briefing ahead of Google I/O 2026, I watched company execs unveil a list of AI-powered features aimed at solving pain points across its software ecosystem. 

One tool promises to radically improve the quality of video searches: Ask YouTube, as it’s called, scours the platform’s catalog of long-form videos and Shorts to surface content relevant to more complex queries. 

At first glance, that sounds like a win for both YouTube viewers and creators. Ask YouTube, however, takes an extra step — it directs searchers to a video that answers their query and zeroes in on the relevant timestamp. Get in, get your answer, get out. 

But if people don’t stay to watch an entire video, or at least most of it, this threatens the ways video creators make money and build a following. YouTubers need a large subscriber base, which leads to more ad revenue, sponsorships, affiliate links and fan funding. 

In other words, Google’s pursuit of user convenience could directly defund the creators who power its platform. Goodbye, ad revenue. So long, community building.

Google made a similar move when it introduced AI Overviews in late 2024, ranking readers over publishers that had long relied on traffic from the search engine results page. By scraping information from media sites and summarizing it at the top of search results, AI Overviews reduce clicks to other websites by 58%, according to a February report by marketing and research firm Ahrefs. (Google claimed that “links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query.”)

In the opening minutes of last year’s Google I/O 2025 keynote, Google CEO Sundar Pichai bragged that AI Overviews — a feature that can’t be turned off and automatically appears — had over 1.5 billion users per month. Alongside today’s Google I/O 2026 keynote, Pichai penned a blog post noting that the feature had grown to over 2.5 billion users per month.

The cannibalization of referral traffic by Google’s generative AI Overviews appears to be a predictive model for video platforms. The Ask YouTube feature, as described, will direct you to the point in a video answering their question, after which you’ll likely have little or no incentive to stay. Viewers likely won’t get a sense of the channel’s subject and vibe, nor will they keep watching to engage with its charm or storytelling. 

Google’s “solution” is to surgically excise creator content to funnel YouTube viewers to answers. In essence, it would replace YouTubers’ revenue streams by harvesting their data and expertise, potentially undermining their entire business model.

Ask YouTube is initially coming to YouTube Premium subscribers, but the company plans to roll it out to the whole platform at some point. 

Coding Search answers on the fly — instead of finding the right YouTube video

Google introduced endless AI features during its I/O keynote, but some are more subtle in their efforts to replace content creators’ revenue streams. Search has some integrations with intriguing capabilities, including using the new Gemini 3.5 Flash’s AI agent capabilities to code simple, quick software on the fly. For example, from a basic request, it can build a wedding or travel-planning widget that surfaces relevant info and deadlines right in the browser window.

But this so-called agentic coding has another use case that Google demonstrated in the briefing. Say you asked a question about a specific part of astrophysics, like black holes — to answer, Gemini 3.5 Flash could craft an interactive visual simulation to demonstrate how an elaborate concept works. Google calls it “generative UI.”

I would hope that most people seeking to understand black holes would turn to a reputable publication with high-quality space and science journalism. But realistically, I’d expect many folks to head to YouTube to see a complex astrophysics concept broken down visually in a well-produced video. Why would they bother if Gemini 3.5 Flash whips up a simulation straight from search results? 

There’s a lot of uncertainty, especially around the quality and precision of simulated answers generated by the new Gemini model. Yet even if they’re occasionally inaccurate, the convenience seems to outweigh their trustworthiness. 

Despite AI Overviews telling people to eat rocks back in 2024 and providing other misinformation, people still use it instead of scrolling down the page. The AI startup Oumi, quoted by The New York Times, estimated that AI Overviews using the newer Gemini 3 model were accurate 91% of the time. A handful of health care organizations and charities told The Guardian that AI Overviews gave inaccurate suggestions for searches on pancreatic cancer, liver disease and other serious health conditions.

I wouldn’t be surprised if the explosion of new AI functionality in Search leads to fewer people looking up answers in videos. Search can’t replace streamers (yet), but a whole category of YouTube content — explainers, how-tos and walkthroughs — could see a drop in traffic if viewers don’t venture beyond their AI-generated search query results. And they’ll miss out on all the curious, weird, wild people making videos to share with the world.

As AI Overviews reduce clicks outside Google’s ecosystem, YouTube’s conversational AI and Google’s generative search simulations threaten to shrink video content production. And that creates a paradox: If creators stop making informative content due to a lack of traffic and compensation, AI models won’t have the datasets they need to generate future answers. 

Ultimately, Google is building its empire of “convenience” on the back of uncompensated creators.