AI Search Is Ahead of AI Shopping. A New Survey Shows How Far

Generative AI is already a search channel for many of the people using it. Autonomous shopping remains a smaller behavior, and consumers still want a trusted brand in the transaction.

That is the clearest read from Channel V Media’s report updated October 2 with June data from Prosper Insights & Analytics. The survey covered 7,675 U.S. adults. Among people who use generative AI, 53% said they use it to search the internet, making search the most common listed task. Only 7.7% of all respondents said they use agentic AI today.

The two percentages use different denominators. The search figure is among generative AI users; agent use is among all adults. Channel V estimates that about 21.5% of all U.S. adults search with AI. That matters when marketers turn a consumer survey into a media plan.

Willingness and behavior diverge. 40.1% of respondents said they would let an AI agent handle at least one of 13 everyday tasks, but only 17.6% thought agents were a good idea. Channel V says 41.8% rejected the idea and 40.6% were unsure. The willingness figure reflects a stated response after respondents saw a definition; it isn’t observed purchasing behavior.

This result updates a different consumer survey already covered by The State of AI Marketing in August. Croud reported that 69% of its U.S. respondents were open to letting AI buy for them without approval. The questions, definitions and samples differ, so the percentages aren’t a trend line. Both studies point to a measurement problem: what people say they might authorize and what they do now are separate signals.

For marketers, the useful near-term move is to make brands easy to find and describe inside AI search. People are using these tools to research and compare before they delegate a purchase. That puts accurate product information, useful independent coverage and clear answers in the discovery path, even when the final transaction still happens on a brand’s own site or app.

Channel V president Gretel Going framed the change this way: “Earned media has always been about reaching an audience; now it also decides what AI says about you.”

The agency specializes in public relations and AI visibility, and it selected the questions and interpreted Prosper’s data. That commercial interest belongs beside the finding, especially because the report’s recommendation for brands overlaps with Channel V’s services.

The gap between discovery and purchase also shows up in Expedia Group’s separate YouGov survey, fielded across the U.S., U.K. and India in March. 68% preferred booking with a trusted travel brand over an AI chatbot or agent. Another 66% said they wouldn’t trust an AI assistant to buy or book for them.

Expedia’s Chief AI & Data Officer Xavi Amatriain summarized the barrier: “Travelers don’t have a technology problem with AI. They have a trust problem.” That study is travel-specific and doesn’t establish the same attitude in every category.

The difference matters operationally. AI search can influence which products make a shortlist without closing a sale. A brand’s pricing, availability, reviews and policies need to be clear enough for an AI system to summarize correctly. The brand still needs a useful destination for the person who wants proof, service or control.

One risk is overreading this as a call to show up in every answer engine. Channel V’s study is a single June survey, not a panel tracking the same people over time. Its “agentic AI” measure also depends on a definition respondents saw during the questionnaire, and it can’t show which marketing activity causes a brand to surface in a generated answer.

Pew Research Center’s September 30 study found synthetic survey respondents often missed human opinions, a reminder that even sophisticated AI research can flatten real variation. That finding concerns synthetic polling, not the consumer behaviors reported here. Other SOAM coverage has already examined how willingness to let AI buy differs from shoppers’ validation habits and why product feeds shape agent commerce.

The report’s strongest practical signal is narrower: AI-assisted search is already ahead of AI delegation. Teams should track whether the brand appears when customers research a category, then measure what those people do afterward. Agent-ready checkout can stay on the roadmap. The nearer-term job is earning a place in the shortlist and a reason for the shopper to trust the next click.