Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Shoppers who use AI are paying a hidden verification tax

Why consumers need a system to verify AI answers.

[Photo: Getty Images]

Profile photo of Michael Quoc

Founder & CEO · Product.ai

In theory, artificial intelligence should make shopping faster and easier. In reality, shoppers are getting burned by AI answers they can’t rely on. As a result, shoppers now spend more time on research because they can’t trust what they’re told. Instead of accepting AI product recommendations as valid, most people double-check them by reading reviews and asking friends, family, and social media contacts. The shopping experience has become an investigation.

Product.ai’s 2026 Trust in AI Commerce Report, a survey of 1,463 U.S. online shoppers that my company fielded in April, highlights the new consumer experience: 86% of the 623 respondents who use AI for product research said they confirm its recommendation through another source before they buy. Similarly, in Reddit’s Path to Purchase 2026 Survey of nearly 14,000 U.S. shoppers, half said they verify AI recommendations on Reddit before buying.

After years of building commerce systems, I’ve never seen consumer behavior change so fast.

AI HAS MADE SHOPPING MORE DIFFICULT

I call this extra research to catch AI’s mistakes the “hidden verification tax.” It’s the time and energy spent scanning comparison charts, getting second opinions, and scouring reviews, all of which are now a normal part of today’s shopping experience. When the Interactive Advertising Bureau screen-recorded more than 450 real shopping sessions, it found that shoppers who consulted AI took 3.8 steps before buying, up from 1.6 steps before AI entered the journey.

The problem is that AI answers have added more untrustworthy noise to shopping deliberations. Our Trust in AI Commerce Report asked people to rate seven common sources of buying information on a 10-point scale. Friends and family came in first at just 5.13. AI placed sixth at 4.51. Even the top source barely cleared half.

THE TRUST FACTOR

Many people don’t trust AI with product recommendations. The newest behavioral research suggests the verification tax isn’t going anywhere soon. A 2025 meta-analysis in Psychological Bulletin, spanning 163 studies and more than 82,000 participants, found that people distrust algorithmic advice by default and warm to it only when the AI is clearly more capable and the decision doesn’t feel personal. Because shopping is often a deeply personal experience, it’s no surprise that a 2025 study in the International Journal of Consumer Studies also found that consumers trust people more than AI for open-ended tasks that involve taste.

Julian De Freitas at Harvard Business School and his co-authors organized resistance to AI into five psychological barriers. Two of the barriers are directly relevant to shopping. The first is opacity: AI is a black box, so you never see why it picked one product over another. The second is emotionlessness: AI is unfeeling, so it can’t know your taste and the emotions behind your purchase.

WHY SHOPPERS KEEP USING AI ANYWAY

Paradoxically, distrust hasn’t slowed adoption, perhaps because we don’t fully realize that verifying is costing us time. Adobe Analytics reported that traffic to U.S. retail sites from generative AI tools rose nearly 700% year over year during the 2025 holiday season. In our Trust in AI Commerce Report, 43% of shoppers surveyed had used AI for product research in the past 90 days.

But all this research still leaves doubts behind. In our study, 15% of shoppers wished they had spent more time researching their most recent purchase of $50 or more. And 13% found a better option after buying.

Shoppers’ answer to this purchase-decision worry is to use even more AI. In a recent Harris Poll for Quad/Graphics, Inc., 51% of consumers overall, and 62% of Gen Z and millennials, said they would rather use AI-powered shopping tools to reduce the risk of a bad purchase. The proliferation of AI content itself may have raised that very risk.

WHO PAYS FOR INACCURATE AI ANSWERS

If nothing happens, we’re heading toward a world in which AI answers are taken on faith. Anyone who believes a mistaken or inaccurate response pays for it in one way or another.

The hidden costs of AI reliance are already surfacing in offices around the world. In a University of Melbourne and KPMG study of 48,000 people across 47 countries, two in three employees using AI admitted relying on its output without evaluating it. Over half of respondents had already made mistakes because of their reliance on AI.

THE BETTER FUTURE HAS VERIFICATION BUILT IN

Imagine a different approach that would protect consumers from bad AI answers and save them time—a verification layer that checks AI recommendations against independent, non-branded sources before you act on them.

This kind of verification layer, no matter who builds it, cannot have results that are for sale to the brands it evaluates. And consumers agree. In the Harris Poll, 75% of the 2,180 U.S. adults surveyed said they would trust an AI shopping agent less if its recommendations were swayed by brand dollars.

So, where does that leave us? Consumers will continue to struggle through shopping decisions that require double- and triple-checking product information until someone builds the verification layer they can trust. Whoever automates what shoppers do manually today will own the most valuable layer in AI commerce. When that happens, we will all stop paying the verification tax for good.

If you’re building something similar, or you think the premise is wrong, send me a note. I’d like to hear how you’re thinking about it.

Michael Quoc is founder and CEO of Product.ai.