Two numbers from the same survey don't usually pull in opposite directions this hard. Seventy per cent of consumers say they've used AI tools for search more over the past year, according to a 2026 study. And the share who say AI search is more helpful than a regular search engine has dropped from 82% to 54% in that same year. People are leaning on AI search harder while trusting it less, and the marketers in the same survey are seeing it show up in their own numbers too.
The data comes from a 2026 survey by the marketing agency Fractl, run with Search Engine Land, of 1,008 US consumers and 150 marketers (Search Engine Land, "AI search adoption rises as consumer trust declines: Study"; Fractl, "AI Search Consumer Trust Study: Brand Visibility Strategies for 2026"). It's one survey, not a census, but the size and the split between a consumer sample and a marketer sample make it one of the more concrete reads on this gap so far, and it's worth ten minutes to see what it found.
What the survey found
The headline number, according to the survey, is a 28-point swing in helpfulness ratings: 82% of consumers rated AI search more helpful than traditional search in 2025, and only 54% said the same a year later. That drop shows up alongside, not instead of, rising use. Seventy per cent of respondents said they use AI tools for search more than they did a year ago, and only 3% said less. Put the two together and the picture is less "people are giving up on AI search" and more "people keep using it while trusting it less each time."
The clearest sign of that, according to the same data, is the size of the sceptic group: consumers who rate AI search as less helpful than a traditional search engine grew from 3% in 2025 to 17% in 2026, nearly six times as many. That's still a minority, but it's a fast-growing one, and it grew inside a population that was, on the same survey, using AI search more than ever.
One split in the data cuts against the assumption that younger users trust AI more by default. Baby boomers rated AI search as more helpful (63%) than Gen Z did (47%), according to the survey. The generation with the most day-to-day exposure to these tools came out more critical of them, not less.
Why is trust falling while use keeps rising?
The survey doesn't answer this directly, so anything past the numbers is our reading of them, not a finding of the study itself. Our guess is that the two trends aren't in tension: using a tool more is often exactly how you find its rough edges. A search engine result you can scan and dismiss in a second; an AI answer takes longer to read and is stated with more apparent confidence, so a wrong or dated one costs more of a user's attention before it's caught. The more someone uses AI search, the more chances they've had to notice it get something wrong, confidently.
That would also explain the boomer/Gen Z split. If familiarity is what's driving the scepticism, then a generation that grew up expecting search engines to be right most of the time may simply have noticed the gap faster than a generation with less of a baseline to compare against.
The brand-trust risk: heavy AI use as a liability
The part of the survey most directly relevant to a marketing team isn't about AI search generally, it's about what happens when a brand leans on AI visibly. According to the survey, in 2025, 20% of consumers said that a favourite brand using AI heavily would reduce their trust in it. In 2026, that figure had roughly doubled to 40%.
That shift wasn't even across age groups: according to the survey, 54% of Gen Z respondents said visible AI use by a favourite brand would reduce their trust in it, against 33% of Gen X and 32% of Baby Boomers. The generation most sceptical of AI search overall is also the one most likely to penalise a brand for leaning on AI in its own marketing. Two years ago, a brand talking about "AI-powered" features was mostly making an innovation claim. On this data, for a meaningful and growing share of consumers, it's closer to a liability disclosure.
None of this means AI adoption itself is the problem. It means the old assumption, that visible AI use reads as forward-thinking by default, no longer holds for a large enough slice of the audience to plan around.
What this means for how you show up in AI answers
None of this is a reason to pull back from AI search as a channel; people are using it more, not less, and that alone means it's worth paying attention to. But it changes what "showing up well" in an AI answer is worth, and to whom.
A separate 2026 consumer research report from Yext offers a useful data point here. According to Yext, 52% of people who get a recommendation from an AI assistant click through to a source the AI cited in its answer, rather than taking the answer at face value (Yext, "7 Data-Backed Stats on AI Search Trust and Consumer Decision-Making in 2026"). Read next to the Fractl and Search Engine Land numbers, that's the practical shape of rising scepticism: people don't stop asking AI assistants for recommendations, they just check the working more often than they used to. When they do, according to the same Fractl survey, traditional Google search results are still the single most trusted source for a product recommendation (39%), ahead of Reddit (15%) and AI tools themselves (14%). That's the argument for treating an AI assistant's citations as something worth watching directly, not just the headline recommendation: if someone asks ChatGPT or Perplexity about your category and a chunk of that 52% click through to check, the pages an assistant actually cites, and whether they say what the assistant claims they say, matter as much as whether you were recommended in the first place.
<img src="/aiv/dashboard.png" alt="Babel42's AI Visibility dashboard for a demo email-marketing brand, showing appearance rate and AI win rate across buyer journeys on Claude and Perplexity, share of AI voice broken out by mentions versus by citations, and a top cited sources list led by a review-aggregator site and the brand's own domain" />In our own data above, from a demo account tracking an email-marketing brand, the share of AI voice by mentions runs to 24% against just 10% by citations, a gap worth checking on your own brand in a rising-scepticism environment: showing up in an AI's answer and being one of the sources it names are two different numbers, and the second one is what a sceptical reader clicks through to verify. It's the same discipline we've written about before: what actually earns a citation once you're being measured covers the mechanics, and what AI search visibility measures in the first place is the place to start if the concept itself is new.
Two things worth doing this week
- Look at what an AI assistant cites, not just what it recommends, the next time you check your own category. Ask ChatGPT, Claude or Perplexity a buying question in your space and read the sources it names, not just its final answer. If your own site or a review platform you're active on isn't among them, that's the gap a sceptical, click-through-happy user would find too.
- If your marketing describes a product as "AI-powered," check who that's aimed at before you lean on it harder. On this data, the same phrase reads as a plus to a Baby Boomer and a risk to a Gen Z reader roughly half the time. That's not a reason to hide the feature, but it's a reason to think about which audience is reading the page.
The short version
According to a 2026 survey by Fractl and Search Engine Land, of 1,008 US consumers and 150 marketers, AI search use keeps rising (70% of consumers say they use it more than a year ago) while trust in it falls (the share rating it more helpful than traditional search dropped from 82% to 54% in the same year, and the sceptic share grew from 3% to 17%). Baby boomers now trust AI search more than Gen Z does, and the same generational pattern reverses for brand trust: 54% of Gen Z say a favourite brand leaning heavily on AI would reduce their trust in it, against roughly a third of older generations. According to a separate Yext report, 52% of people click through to an AI answer's cited sources rather than taking the answer at face value. Put together, the practical implication for a marketing team is that being cited, specifically and checkably, is becoming more valuable precisely because fewer people are willing to take an AI's word for it alone.
