Monthly letter · News · Topics · Identity · Dispatches · Est. 2026

Where the Real You Learns to Live, not Just Stand Up

TheSignal

Dispatch · AI & the Self

Disclosing AI use at work: 13 experiments on trust and a preliminary interview study of medical writers' professional identity

University of Arizona researchers ran 13 experiments with more than 5,000 participants on how people judge those who disclose AI use. A conference abstract by authors at PHAROS Labs in Hamburg looked at professional identity among medical writers at a German market access consultancy that mandated AI use. The first measures other people's judgments, the second the writers' own.

James H. TippinsThe Autonomy Commission
DispatchAI & the Self
SourcesUniversity of Arizona News on a study by Reimann and Schilke (May 22, 2025) · ISPOR Europe 2025 abstract, Value in Health 28(S2) (November 2025)

University of Arizona News (Andy Ober) reported on May 22, 2025 on 13 experiments with more than 5,000 participants by Martin Reimann and Oliver Schilke of the Eller College of Management. Disclosing AI use reduced trust in all 13. Students' trust in a professor fell 16% after learning the professor used AI for grading, investors trusted firms 18% less when ads disclosed AI use, and clients trusted graphic designers 20% less. Familiarity with AI did not prevent the drop, and softer framing, such as saying AI only proofread the work, still lowered trust. The page does not name the journal (it gives a DOI), report the sample breakdown by experiment, or include a limitations section.

The ISPOR Europe 2025 abstract (Cuk, Wiehe and Woeste of PHAROS Labs, Hamburg; Value in Health, vol. 28, issue S2) describes interviews with medical writing professionals at one German market access consulting firm where AI tools were mandated for daily use. The planned sample was 10 interviews, but the results cite only three completed ones, and the authors call the findings preliminary. Participants reported 30 to 40% time savings on draft preparation, a self-reported figure. Two identity threats emerged: to professional recognition and to autonomous judgment. The abstract quotes a concern about becoming a generative AI operator rather than a strategic advisor.

Read together, one source records how observers rate someone who says AI helped, and the other records how a few users described what the tool meant for their standing. Neither measures what a worker does when the audience's judgment and their own sense of their role pull in different directions.

◆ My Take

Trust after disclosure and identity under a mandate

The Arizona experiments put the cost on the person who says AI was used: lower trust in every one of 13 tests, even with softer wording. The abstract from the consulting firm describes an inside view in which writers who found AI useful still named recognition and independent judgment as at stake.

The gap between them is wide. The experiments use scenarios with outside evaluators, and the abstract rests on three interviews in one company, so neither reports how a worker weighs outside judgment against their own sense of the role.

◆ If this is your question

Work on it out loud.

Free Mind Thursday is my weekly live call, and it costs nothing. Sign up at beyondunbound.com/freemind. Or get the next issue of The Signal free, below.