‘Let Me Explain’: A Comparative Field Study on How Experts Enact Authority Over Clients When Facing AI Decisions
Citation. Mayer, A.-S., van den Broek, E., & Karačić, T. (2026). ‘Let Me Explain’: A Comparative Field Study on How Experts Enact Authority Over Clients When Facing AI Decisions. Journal of Management Studies, 63(2), 366–398. https://doi.org/10.1111/joms.70022
Summary. This comparative field study examines how experts retain authority with clients after predictive AI takes over decisions that experts must still present and justify. Across banking, biotechnology, and recruitment, experts used explaining practices to help clients understand and accept AI decisions. Whether those practices reconstructed authority depended on clients recognizing human expertise as distinctive and on experts having sufficiently rich interaction to learn and respond to evolving client needs.
Key claims
- When experts could not overrule AI decisions but remained responsible for presenting them, explaining became an alternative pathway for enacting authority with clients rather than a cosmetic communication task.
- Explaining practices were more likely to reinforce expert authority when clients recognized the expert’s unique contribution and when interactions gave the expert enough information to tailor explanations to client needs.
- Where experts failed to incorporate client needs into AI-decision explanations, clients could neither understand nor endorse the outcome, which undermined rather than restored expert authority.
- The cases show that the work left to human experts after AI decision automation includes relational sensemaking and acceptance-building, not merely residual presentation of a technically produced decision.
Evidence & limitations
- Comparative qualitative field study of loan consultants and customers at a German bank, seed sorters and supply-chain managers at a biotechnology firm, and recruiters and senior managers at a consumer-goods firm; it draws on approximately 381 observation days, 139 semi-structured interviews or conversations, and organizational documents collected between 2018 and 2025.
- The design supplies rich process evidence across three settings but does not estimate prevalence or causal effects of explaining practices, client acceptance, or authority outcomes; the cases concern predictive-AI decisions in client-facing expert work and may not transfer directly to GenAI or autonomous-agent workflows.
- The article was published online in 2025 and in the March 2026 journal issue; the Drive PDF is an open-access Wiley article with matching title-page DOI, and the stored binary hash identifies the ingested file.