Confidence as a variable: a Turkish nursing school measures students' trust in their own AI use

A research team at Koç University School of Nursing in Istanbul has published a measurement instrument designed to do something most AI-in-education surveys cannot: capture, on a single validated scale, how confident university students actually feel about using artificial intelligence in their academic and clinical work. The study, led by Associate Professor Remziye Semerci Şahin and her colleagues, was reported on 18 July 2026 and treats self-efficacy, the belief that one can successfully perform a task, as a variable worth scoring rather than assuming.

That distinction matters. Universities across the OECD have spent the past two years debating whether students use AI to cheat, to learn, or simply to keep up. Almost none of those debates rest on a standardised measure of the user's own confidence in the tool. The Koç paper inserts itself into that gap. The argument running through the methodology is direct: until researchers can score how competent students believe they are at handling AI, any claim about how AI is reshaping nursing education is built on sand.

What the scale actually measures

The instrument is described in the source material as a self-efficacy scale, built and tested on nursing students at a single Turkish private university. Self-efficacy scales are a long-standing tool in health-professions research, typically adapted from the work of psychologist Albert Bandura in the late 1970s. The Koç team applied that tradition to a specific modern artefact: a student's confidence in selecting, prompting, evaluating, and trusting AI outputs in a clinical or academic setting.

The practical move is to convert a fuzzy impression ("students seem comfortable with AI") into a numeric score that can be tracked across cohorts, compared between programmes, and correlated with actual performance. The source thread does not publish the full questionnaire, the sample size, or the resulting mean scores. What it does confirm is that the researchers treated self-efficacy as the dependent variable of interest, and that the population studied was nursing undergraduates at Koç. That places the work inside a much larger global conversation about competency-based assessment in health-professions training, where regulators increasingly demand evidence that graduates can handle the digital tools already standard on hospital wards.

Why nursing, and why now

Nursing education is an unusually demanding test case. AI tools are already being integrated into triage chatbots, documentation assistants, and clinical decision-support systems in hospitals from Istanbul to London. A graduate nurse who cannot critically evaluate an AI-suggested intervention is a liability, regardless of how well they handle a manual. The Koç study frames self-efficacy as the bridge between a curriculum that teaches AI and a clinical environment that will assume it.

There is also a structural pressure underneath. Turkey's higher-education sector expanded rapidly over the past two decades, with private universities like Koç competing on technology integration as a recruitment draw. Confidence in AI is, in that sense, also a market signal. Students who feel fluent in AI are more attractive to employers; universities that can document that fluency through validated instruments have a defensible marketing claim. The source reporting does not make this argument explicitly. It does not need to; the choice of population and instrument speaks for itself.

What remains uncertain

A single-site study at a well-resourced private university in Istanbul tells the reader something specific, and not much more. The source thread does not specify whether the instrument has been cross-validated at public universities, at health-science schools in other provinces, or among practicing clinicians returning for graduate study. It does not report effect sizes, item-level reliability statistics, or how scores correlate with actual clinical AI use. None of that is in the available reporting. Until a multi-site replication appears, the scale is best read as a promising, locally validated instrument rather than a generalisable national measurement.

There is also a counter-reading worth naming. Critics of AI-in-education research have long argued that self-reported confidence in using a tool is a poor proxy for competence. Students regularly report high confidence in skills they cannot demonstrate; they also report low confidence in skills they perform adequately. If future replications of the Koç instrument show weak correlation between self-efficacy scores and observed AI use in clinical placements, the scale will need to be re-cast as a measure of perception rather than capability. The Koç team has not yet had to face that test in published form.

The structural question underneath

The deeper pattern here is the slow translation of human-skills measurement into the AI era. For three decades, medical and nursing education has leaned on validated scales for empathy, burnout, clinical reasoning, and communication. Adding AI self-efficacy to that toolkit is overdue rather than novel. What is new is the speed: the tools students are being asked to feel confident about are themselves changing faster than any instrument can be revalidated.

That tension sits underneath the Koç study. A scale published in mid-2026 measures confidence in a category of tools that will look meaningfully different by the time a 2024 cohort graduates. The instrument is a snapshot. The race it joins is not.

Desk note: Monexus framed this as a methodological development, not a productivity claim. Wire coverage of AI in education tends to track adoption rates; the Koç paper is interesting precisely because it tries to measure the user behind the tool.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://en.wikipedia.org/wiki/Self-efficacy
• https://en.wikipedia.org/wiki/Ko%C3%A7_University
• https://en.wikipedia.org/wiki/Nursing_education