Nobody is automatically ready for AI
Three new studies show that readiness for clinical AI is shaped by personality, digital access and career fears far more than by simply growing up with the technology.
We tend to assume that readiness for AI will take care of itself. The current cohort of medical students has never known a world without smartphones, so surely they will absorb clinical AI the way they absorbed everything else, by osmosis and generational fluency. It is a comforting story, and this week’s papers suggest it is largely wrong. This isn’t surprising, as the concept of the ‘digital native’ in education was largely debunked before the popular emergence of genAI1.
I have selected three papers this week, and together they move the conversation from whether students like AI to who is actually prepared to use it well, and why.
Key points:
Personality, not just tech enthusiasm, predicts how ready a student feels. Openness and agreeableness were independent predictors of AI readiness, while a general affinity for technology faded once personality was taken into account.
A persistent gender gap appeared, with men reporting higher overall readiness and women scoring higher on the ethical dimension.
In one large survey, AI use fell in the later, clinical years of training, and students with more AI knowledge reported using it less, a reversal of the usual adoption pattern.
Lack of training and limited technical access were the biggest barriers, a reminder that readiness is a resource question as much as an attitude.
AI is already reshaping career decisions, with a third of radiology-interested trainees saying it has discouraged them from the specialty.
Bornemann-Cimenti et al. ran the kind of study that is surprisingly rare, a single multinational cohort that measured personality, technology affinity and AI readiness at the same time, published in BMC Medical Education2. Drawing on 1,278 medical students from six continents (out of 1,920 who began the survey), they paired the Medical Artificial Intelligence Readiness Scale with the Big Five Inventory and an affinity-for-technology scale.
The headline is that psychology matters more than gadget enthusiasm. Openness tracked closely with the forward-looking “vision” side of readiness (r = 0.67) and agreeableness with ethical awareness (r = 0.60), and in regression, openness and agreeableness stayed as independent predictors of overall readiness while a general affinity for technology dropped out once personality and gender were included. There was also a clear gender signal, with men reporting higher overall readiness and women scoring slightly higher on ethical awareness.
The design is cross-sectional, so these are associations rather than causes, but the practical implication is useful even so. If readiness is bound up with stable traits rather than simple exposure, a single generic “AI in medicine” lecture will land very differently across a cohort, and curricula may need differentiated approaches that meet sceptical or anxious students where they are instead of assuming enthusiasm is universal.
Qazi et al. surveyed 501 medical and dental students in Khyber Pakhtunkhwa, Pakistan, in Medical Education Online, and found a pattern that should give curriculum designers pause3. Knowledge and attitudes were reasonably positive, but actual use was modest, and it did not climb with seniority. Second-year students reported more AI use than fourth-years and final-years, and being in fourth year roughly doubled the odds of low use, an association that strengthened to nearly fourfold in sensitivity analysis. Strangest of all, higher knowledge predicted lower use, the opposite of the familiar “know more, use more” curve.
The authors are careful that the data are cross-sectional and exploratory, and they offer a professional-identity reading: as students move onto the wards and begin to feel like clinicians in training, they may exercise deliberate restraint around tools they do not yet fully trust. The barriers were prosaic and important, with 71.7% citing a lack of training and 69.7% limited technical access.
Read next to the multinational study, the message is that readiness is not only psychological but material. In lower-resourced settings especially, it depends on infrastructure, on faculty who can teach it, and on ward cultures that allow it.
Yilmaz et al. bring the consequences into focus by looking at how AI is shaping specialty choice, in a national survey of 401 Canadian medical students and residents published in Academic Radiology4. AI had at most a modest influence on most people’s specialty decisions, but the effect concentrated where you would expect, among those considering radiology, and it skewed negative.
Roughly a third (32.9%) of radiology-interested respondents said AI had discouraged them from the field. The most revealing twist concerns the students who understood AI best. Those the authors classed as “AI-savvy” were more likely to see AI as a tool that augments rather than replaces radiologists, yet they were also more likely to express hesitation about committing to the specialty, apparently because deeper knowledge brought sharper awareness of potential disruption.
Notably, believing that AI improves medical education was not associated with interest in radiology, so the anxiety was about the job, not the teaching. For educators this is a warning that our AI messaging does not stay in the lecture theatre. What we tell students about AI’s trajectory feeds directly into workforce decisions, and loose talk about machines replacing whole specialties may steer capable students away from fields that will still very much need them.
The throughline is that readiness is made, not inherited. It grows out of personality and access and mentoring and the stories we tell about where the technology is heading, and none of those arrive automatically with a new generation.
The most useful shift these papers invite is to stop treating readiness as a milestone students will pass on their own and to start treating it as something we build, unevenly distributed today and easy to erode with the wrong curriculum or the wrong messaging. That is a demanding reframing, because it places the responsibility on how we teach, but it is encouraging for exactly the same reason. If readiness is something we construct, then it is something we can learn to do better.
Bornemann-Cimenti, H. Personality traits, technology affinity, and artificial intelligence readiness in medical students: a multinational cross-sectional study. BMC Med Educ(2026). https://doi.org/10.1186/s12909-026-09869-2
Qazi, S., Khan, I. A., Waleed, M., Ahmad, A., Qarni, O., Abdullah, M. W., … Mazhar, M. A. (2026). Artificial intelligence readiness in Pakistan’s medical and dental education: training-phase decline, a knowledge–practice paradox, and the role of digital determinants of health. Medical Education Online, 31(1). https://doi.org/10.1080/10872981.2026.2673253
Yilmaz E, Atalla M, Gaisinsky A, Chan A, Lin C, Damer A, Fleming R, Bilbily A, Tyrrell PN. The Impact of Artificial Intelligence on Radiology Specialty Preferences Among Canadian Medical Students and Residents. Academic Radiology (2026) https://doi.org/10.1016/j.acra.2026.04.039