68/100

Moderate evidence

Cite
More
EmailLinkedInX

Personalised AI recommendations won more trust and acceptance in four studies, with gaps between ethnic groups

Study title: Different Strokes for Different Folks: Exploring Algorithmic Fairness and Trust in Generative Artificial Intelligence

Evidence type
Controlled or validated comparison
Sample
n = 776
Sources
1 sourcePeer-reviewed
Published
1 min read

Study title: Different Strokes for Different Folks: Exploring Algorithmic Fairness and Trust in Generative Artificial Intelligence

A peer-reviewed paper in Psychology & Marketing reports four linked studies on how personalised generative AI recommendations are received by consumers. In the first, an experiment with 136 participants, highly personalised recommendations led to greater acceptance, higher perceived fairness and more trust than low personalisation.

Read the study abstract

ABSTRACT Generative artificial intelligence (GenAI) is widely used to personalize recommendations, yet questions remain about the representativeness of algorithmic recommendations to recognize diverse consumer needs. This research integrates the stimulus‐organism‐response model and social identity theory to inform how personalization influences recommendation acceptance through algorithmic fairness and trust and how communication style affects these processes, which may differ between ethnic groups. Using a sequ…

The gain

In a 136-person experiment, highly personalised generative AI recommendations increased recommendation acceptance, perceived algorithmic fairness and trust compared with low personalisation.

In brief

A second experiment with 300 participants found that personalisation raises acceptance through a chain running from perceived algorithmic fairness to trust. Communication style mattered: competence-oriented messaging produced a stronger effect than warmth-oriented messaging.

A quasi-experiment with 340 participants found these relationships differed across ethnic groups, and 14 semi-structured interviews were used to explore the reasoning behind the results. The authors say the findings point to more equitable, context-sensitive design for diverse consumers.

Main points

  1. Study 1 (136 participants, experiment): high personalisation increased recommendation acceptance, perceived fairness and trust versus low personalisation.
  2. Study 2 (300 participants, experiment): a positive sequential indirect effect ran through algorithmic fairness and then trust.
  3. Competence-oriented communication produced a stronger effect than warmth-oriented communication in Study 2.
  4. Study 3 (340 participants, quasi-experiment): the relationships differed across ethnic groups.
  5. Study 4 (14 participants): semi-structured interviews were used to contextualise the findings and explore underlying reasoning.
  6. The authors frame fairness and trust as key pathways to recommendation acceptance and recommend more equitable, context-sensitive GenAI design.

The full story

The research combines the stimulus-organism-response model with social identity theory. It asks how personalisation influences whether people accept generative AI recommendations, and how perceived fairness and trust shape that link.

It used a sequential explanatory design. Studies 1 and 2 were experiments with 136 and 300 participants. Study 3 was a quasi-experiment with 340 participants, used to compare ethnic groups.

Study 4 added 14 semi-structured interviews to explore the reasoning behind the quantitative results. The authors say the findings can guide GenAI developers toward more equitable and context-sensitive design for diverse consumers.

Sources

1 on file
  1. Psychology & Marketingdoi:10.1002/mar.70306
    Peer-reviewed2026-11-17

Talk it over

Three questions to start on, two more if you have time
In Study 3, a quasi-experiment with 340 participants, the relationships between personalisation, fairness and trust differed across ethnic groups. Who might be missing from that comparison, and how could that change what the result means?

Start with a question

Personalised AI recommendations won more trust and acceptance in four studies, with gaps between ethnic groupshttps://truvace.com/item/different-strokes-for-different-folks-exploring-algorithmic-fairness-and-trust-i

  1. In Study 3, a quasi-experiment with 340 participants, the relationships between personalisation, fairness and trust differed across ethnic groups. Who might be missing from that comparison, and how could that change what the result means?

    Answer in the thread
  2. Study 2 found competence-oriented messaging had a stronger effect than warmth-oriented messaging. What finding in a real shopping setting would change your mind about that result?

    Answer in the thread
  3. Study 4 used 14 semi-structured interviews to explore the reasoning behind the numbers. If you were one of those interviewees, what would you want the researchers to ask about fairness and trust?

    Answer in the thread
If you have more time
  • The authors call for more equitable, context-sensitive design. Which groups of consumers would you ask developers to test with first, and what would you want them to measure?

    Answer in the thread
  • Study 3 was not randomised, so the ethnic-group differences could have other causes. What else might explain them, and how could a later study check?

    Answer in the thread

What readers say