Marketing in 3 · Issue 15 · Week of 20 April 2026
When they expect to be judged for who they are. In ten US studies, Black and other non-White customers leaned towards a machine instead of a person whenever the people around them were all White. At one bank, half of non-White men picked the kiosk when the staff were all White, against just over a third when the staff were non-White too. White customers didn't budge either way. Signs that a business values diversity took most of the effect away.
Ask people whether they'd rather be served by a person or a machine and most pick the person. Yeseul Kim (University of Bath), Dipayan Biswas (University of South Florida), Courtney Szocs (Penn State) and Khalia Jenkins (American University) wanted to know who the exceptions are, and when.
Their starting point is uncomfortable but well documented: in US banks, shops and rental offices, Black customers often wait longer, get less warmth and face more suspicion. Being the only non-White person in the room can bring all of that to mind. The question was whether, in that moment, a machine starts to look safer than a person.
The team ran ten studies. Two were real job adverts with paid work at the end, one was a Facebook ad campaign, and the rest were online experiments set in banks, rental offices and an investment firm. Every time, people chose between a human and an algorithm: an AI interviewer, a robot teller or a digital kiosk.
The first job advert, posted at historically Black colleges, set the pattern. A second run, shared through Black community networks, showed an even steeper jump: from 15% to 37% choosing the AI interviewer. Online, Black participants looking at a bank full of White customers chose the robot teller 38% of the time; in a mixed crowd, 22%.
What tipped the balance was a specific worry. When non-White participants were told they felt out of place at a bank, their concern that staff would size them up by race rose from 2.57 to 3.79 on a 7-point scale, and that concern explained the move towards the robot.
It isn't just people wanting staff who look like them. White men in the bank study said they felt less similar to non-White employees, yet they went to the kiosk no more often (38.8% against 37.5%). Feeling different only mattered for people who had reason to expect unfair treatment.
| Group | Value (%) |
|---|---|
| Non-White men, all-White staff | 50.0 |
| Non-White men, non-White staff | 36.1 |
| White men, all-White staff | 38.8 |
| White men, non-White staff | 37.5 |
Study 3a: 805 US men shown a bank whose staff were all White or all non-White.
The Facebook ads ran for one day at a time in mostly Black and mostly White US neighbourhoods. Where most residents were Black, a diverse advice team did almost as well as the chatbot. Where most were White, the ranking flipped: the all-White team drew the most clicks and the chatbot fell behind. One image, two very different audiences.
| Group | Value (%) |
|---|---|
| Black areas: chatbot | 0.63 |
| Black areas: diverse team | 0.58 |
| Black areas: White male team | 0.3 |
| White areas: chatbot | 0.45 |
| White areas: diverse team | 0.44 |
| White areas: White male team | 0.73 |
Study 3c: one-day Facebook campaigns in mostly Black and mostly White US neighbourhoods.
The pull towards machines weakened wherever judgment felt less likely. Applying to lease a flat, people who had just recalled a time they stood out as a minority chose an AI interviewer 52% of the time, against 36% in the control group. Applying to a social club for people with mixed interests, that same reminder made no real difference.
Telling people the bank was minority-owned did something similar. The jump towards the robot teller shrank to eight points (24% to 32%), against twenty-two (30% to 52%) without that line. And at the investment firm, a short diversity statement was enough to cancel out the effect of an all-White team.
The authors studied race and ethnicity in the US, and they suggest that other groups who expect to be judged, by sexual orientation or by wealth for example, may react the same way. Going a step further than the paper, the same logic could hold at a GP's front desk, an insurance claims line, a debt advice service, a car dealership or a luxury boutique: anywhere a customer might feel assessed as a type rather than helped as a person.
It also changes how to read your channel data. If one group of customers sticks to the app, the kiosk or the chatbot, convenience may not be the whole story. They may be telling you something about the counter.
Marketing in 3 steps
Generally no, but the gap narrows when the staff are all White. In one online bank study, the robot teller's share rose from about one in five to nearly two in five once the other customers in the branch were all White (Kim et al., 2026, Journal of Marketing).
To avoid being stereotyped. Feeling out of place raised worries that staff would judge them by race, and those worries, not a liking for technology, explained the switch.
No. White men noticed the difference just as clearly, yet stuck with the human teller, so the effect depends on expecting unfair treatment, not on difference alone.
Not necessarily. The authors cite cases where lending chatbots rated Black borrowers as riskier. Customers may still see AI as less judgmental, which is a perception to respect, not a guarantee.
Offer a self-service or AI route for sensitive tasks and show clear signs of inclusion, such as a diversity statement. Both reduced the pull towards machines in the studies.
Journal of Interactive Marketing · 2026
Weihrauch, A., & Xie, C.
Four experiments show that advertising AI as "better than humans" can backfire by stirring fears about AI threatening humanity, especially among people already anxious about AI. A short reassuring message alongside the claim softened the reaction. Bragging rights for your AI come with a cost, so add the reassurance.
Journal of the Academy of Marketing Science · 2026
Ma, X., Jin, F., & Dai, X.
Across eight studies, buyers and sellers preferred luxury goods authenticated by human experts over AI, because they felt human experts see the item the way they do. The preference weakened when the purchase was less about social status, the product was not positioned as luxury, or the branding was discreet. Status shoppers want a human eye on the goods, even when the machine may be just as good.
Kim, Y., Biswas, D., Szocs, C., & Jenkins, K. (2026). Human or AI? Identity salience enhances historically marginalized (but not privileged) consumers' choice of algorithmic service providers. Journal of Marketing. https://doi.org/10.1177/00222429261444284
This brief is written for practitioners from the published article. Numbers are as reported by the authors; read the article for full methods.
More issues on AI · Service · Retail
Shoppers leaving a staff-free store pledged about 40% fewer volunteer hours than those leaving a staffed one. Why it happens, and three fixes.
An AI bargainer earned 39% of the pie against 33% for humans and seemed more human than people, yet every bot hurt how people felt about the deal.
People who read bad reviews before listening paid $17.29 to see a show, against $20.45 for people who read none. The review changes the experience, not just the purchase.
Bruno Schivinski is Associate Professor of Marketing at Gdańsk University of Technology. He studies how consumers engage with brands in digital and social media.
