Writing "AI" on your product makes you sell less

Over the last two years the word "AI" has shown up everywhere: on TV boxes, on washing machine spec sheets, on app buttons, on the "powered by AI" badges at the top of landing pages. The logic seems obvious. Artificial intelligence is the technology of the moment, so putting it in the window should make a product look more modern and more desirable.

Research says the opposite. When a product description mentions artificial intelligence, people are less likely to buy it. Not because the product is worse, but because that word, on its own, lowers trust.

In this article we look at the experiment that proved it, the latest surveys, the psychology behind it and, above all, how to talk about AI on your website without scaring customers away.

The experiment: same description, one extra word

In 2024 a team of researchers led by Mesut Cicek, a marketing professor at Washington State University, published an experiment in the Journal of Hospitality Marketing & Management that is as simple as it is surprising.

More than 1,000 American adults were split into two groups. Both read the same descriptions of products and services, with the same features and the same information. The only difference: for one group the description mentioned artificial intelligence, for the other it did not.

The test was repeated across eight different categories, from electronics to services. The result was the same in all of them: the versions with the word "AI" produced lower purchase intention. According to Cicek, including these terms in product descriptions is simply a disadvantage (heise).

"Light" products and "risky" products

The effect is not the same for every product. It is much stronger when a wrong choice is expensive: high-end electronics, medical devices, financial services, self-driving cars. Here the word "AI" made participants worried and uncertain (TechRadar).

With low-risk products, like a robot vacuum, the reaction was milder. But even there, the "AI" version still lagged behind the one without it.

The rule is simple: the more customers have at stake (money, health, safety), the more the word "AI" scares them.

A note of honesty: the study measures stated purchase intention, not actual purchases, and the sample is American. But, as we will see, more recent surveys, European ones included, all point in the same direction.

Why the word "AI" is scary

The Washington State University researchers point to a specific culprit: emotional trust. Not rational trust ("I believe it works"), but gut trust ("I feel comfortable buying it"). When the word "AI" appears, that is what drops, and the desire to buy drops with it.

Behind that drop there are at least four well-documented psychological mechanisms.

1. The black box

We understand how a consultant, a shop assistant or a mechanic works. An algorithm, not so much. If I don't know how a machine decides, I don't know whether I can trust its decision. On an expensive product, that uncertainty is enough to make me close the page.

2. We forgive people, not machines

In 2015 three Wharton School researchers, Berkeley Dietvorst, Joseph Simmons and Cade Massey, gave this phenomenon a name: algorithm aversion. In their experiments, people stopped trusting an algorithm after seeing it make a mistake, even when it made fewer mistakes than a human. We give a person a second chance; a machine, much less so.

3. "My case is different"

In 2019 Chiara Longoni, Andrea Bonezzi and Carey Morewedge studied resistance to medical AI (Journal of Consumer Research). Patients preferred a human doctor even when the AI was just as accurate. The reason: they felt a machine could not grasp what made their case unique. The same thought kicks in when people see an "AI-generated diet" or an "AI-calculated quote".

4. Hype fatigue

Then there is a more recent factor: we are tired of hearing about it. In research by The Harris Poll for 4As and Infillion, presented at the Cannes Lions festival in June 2026, 54% of consumers say they hear so much about AI that it annoys them. 65% would like brands to stop mentioning it for no reason (Marketing Brew). When a word is used everywhere, it stops signalling innovation and starts signalling marketing.

Not an isolated case: the 2026 data

Two years after the study, surveys tell the same story. In every one of them, the people who lose trust when they see AI outnumber those who gain it.

Klaviyo's figure is the starkest: for every customer who trusts a brand more because it visibly uses AI, more than four trust it less.

And the trend is getting worse. According to Capgemini, trust in AI-generated content fell from 73% to 55% in two years (MarTech). In the Harris Poll research mentioned above, 63% of consumers say they are less likely to buy from a brand that uses AI-made ads. That last figure should be taken with caution, because the article reporting it does not give the sample size.

The paradox: people like it, until they know it's AI

One experiment makes all this even clearer. Bynder, a company that manages content for brands, asked 2,000 British and American consumers to read two articles on the same topic. One was written by ChatGPT, the other by a professional copywriter. No labels.

Among those with a preference, 56% found the AI-written article more engaging. But when readers were told a piece of content was AI-generated, 52% said they felt less engaged (MarTech).

It is the same mechanism as the Washington State University study, seen from the other side. The content doesn't change; the label does. And the "AI" label carries all the doubts we have seen: the black box, the fear of mistakes, the feeling of being treated like a number.

So the problem is not using artificial intelligence. It's how you present it.

The UX lesson: sell the result, not the technology

Marketers often quote an idea attributed to Harvard professor Theodore Levitt: people don't want a drill, they want a hole in the wall. AI works the same way. Nobody wants "artificial intelligence": they want to save time, make fewer mistakes, find the right thing straight away.

It is a matter of framing: the same information, presented in a different frame, triggers opposite reactions. It is also a matter of cognitive load. As Steve Krug teaches in Don't Make Me Think, every word that forces users to stop and wonder is an obstacle. "AI" raises questions: what exactly does it do? Can it get things wrong? Does it use my data? A result, on the other hand, is understood instantly.

Here is how microcopy changes when you shift the focus from technology to benefit. These are examples to adapt to your own case.

In every case the technology is the same. What changes is the promise you read first.

Practical guide: 7 rules for talking about AI without losing customers

This doesn't mean hiding artificial intelligence. It means putting it in the right place, with the right words.

1. The result in the headline, the technology in the details

Headlines, subheadings and buttons should talk about what the customer gets. AI can appear further down: in the spec sheet, in a "How it works" section, in the FAQ. Those who want to know will find it; those who don't ask won't be put off.

2. The more customers risk, the less you emphasise it

If you sell something expensive, or something that affects health, money or safety, don't put AI in the shop window. Put guarantees, human oversight and the option to go back first. On a cheap, low-risk product you can afford more, but don't make it the main message.

3. Show the person behind the machine

Emotional trust is built by people. "Reviewed by a consultant", "checked by your coach", a name and a face next to the result: these details answer the fear of the black box and the need to feel like a unique case.

4. Be transparent when AI talks to customers

If your site has a chatbot, say so clearly and always offer a way to talk to a person. People want to know: according to Emplifi, cited by eMarketer, 91% expect brands to disclose when they use AI in marketing. Hiding AI and getting found out is the fastest way to lose the trust you are trying to build.

5. Proof, not adjectives

"Revolutionary", "smart", "next-generation": these words prove nothing. Concrete numbers, reviews, real examples and a clear guarantee are far more convincing than any technology label.

6. Leave users in control

A preview, the option to edit the result, an undo button. Control reduces perceived risk, which is exactly what the word "AI" pushes up. It is one of Jakob Nielsen's 10 usability heuristics: user control and freedom.

7. Test the word

Every audience reacts differently. If you have enough traffic, run an A/B test: the same page, one version with "AI" in the headline and one without. Measure conversions and contact requests. Often, as with the $300 million button, changing a few words is enough to see very different numbers.

The best technology is the one you don't see

This research reminds us of a principle good design has always known: people don't buy technology, they buy results. A buzzword can grab attention, but it doesn't build trust. And without trust there is no purchase.

When artificial intelligence works well, it behaves like any well-designed interface: it disappears. Users don't notice the algorithm; they only notice that everything is faster, simpler, more accurate. The job of whoever designs a website is not to tell customers how advanced the product is, but to make them feel they are in good hands.

Frequently asked questions

Should I remove the word "AI" from my website?

Not necessarily: move it. Take "AI" out of headlines, buttons and prominent badges, and put it where interested people look for it, like the spec sheet or the FAQ. Keep the concrete benefit up front.

Does the effect apply to every product?

In the Washington State University study the effect showed up in all eight categories tested. It is much stronger for high-risk products (expensive electronics, health, finance) and weaker for cheap, low-risk ones.

Why do customers trust artificial intelligence less?

Because emotional trust drops. An algorithm is a black box, we forgive its mistakes less than human ones, and we fear it won't understand our specific case. On top of that there is hype fatigue: according to The Harris Poll, 54% of consumers are annoyed by how much AI is talked about.

If I use AI, do I have to disclose it?

When customers interact with AI directly, for example through a chatbot, yes: transparency is what people expect. According to Emplifi, 91% want brands to disclose their use of AI in marketing. Disclosing it doesn't mean making AI the centre of your message.

Sources

1. Washington State University, Using the term "artificial intelligence" in product descriptions reduces purchase intentions, July 30, 2024. Study by Cicek et al. in the Journal of Hospitality Marketing & Management.
2. heise online, Study: AI in the product description arouses mistrust among customers, 2024.
3. TechRadar, AI a turn-off? Tech customers are apparently already getting bored of smart devices, August 4, 2024.
4. Dietvorst, Simmons, Massey, Algorithm Aversion: People Erroneously Avoid Algorithms after Seeing Them Err, Journal of Experimental Psychology: General, 2015.
5. NYU Stern, Patients are resistant to medical AI and prefer human medical providers, on Longoni, Bonezzi, Morewedge, Journal of Consumer Research, 2019.
6. Marketing Brew, Harris Poll: AI fatigue and less trust in AI-generated ads, June 24, 2026.
7. eMarketer, Shoppers aren't impressed by AI-generated marketing, May 1, 2026 (Klaviyo and Emplifi data).
8. MarTech, Consumers like AI content until they know it's AI, June 1, 2026 (Bynder, Validity, Capgemini and YouGov data).
9. Bynder, Human Touch survey.
10. Nielsen Norman Group, 10 Usability Heuristics for User Interface Design.