Dental students want AI in the curriculum, but they also want to be taught how to challenge it
A mixed-methods study at Queen Mary University of London found a clear appetite for artificial intelligence in dental education, but not as an uncritical add-on. Among 86 survey respondents, 73.3% said AI was inadequately integrated into the curriculum, while 76.7% wanted structured teaching on the topic. The numbers point to a cohort that expects AI to become part of routine clinical work and does not want to enter practice unprepared for it.
The enthusiasm is not in doubt. More than four in five respondents, 86.1%, believed AI would be important in their future professional practice. Yet 65.1% reported a lack of formal training. That gap matters because dentistry is a discipline where algorithmic output can influence diagnosis, treatment planning, radiographic interpretation, triage, and administrative decision-making long before a student graduates. If future clinicians learn only how to use AI tools, without learning how to interrogate them, the risk is not dependence alone. The deeper concern is misplaced confidence.
One sentence captures the tension in the study: students want the technology, but they do not trust their current education to prepare them for it.
The mixed-methods design gives the findings extra weight because the survey data were paired with student focus groups. That combination suggests the demand for AI education is not a passing preference but a broader educational concern. The students were not asking for novelty for its own sake. They were asking for a curriculum that reflects how quickly AI is entering clinical environments and how unevenly its outputs can behave when faced with incomplete, biased, or ambiguous information.
In practical terms, the issue is not whether dental students will encounter AI. They already are. The issue is whether education is teaching them to treat AI as a decision-support tool that requires scrutiny, or as a shortcut that can be accepted at face value. In clinical training, that difference shapes professional behavior from the outset. A student who understands uncertainty will be more likely to question a flawed output, compare it with clinical findings, and accept personal responsibility for the final decision. A student who is taught only operational use may be less prepared to notice when an algorithm is wrong in a patient-specific context.
The study also points to a broader regulatory concern. Separate research commissioned by the General Dental Council reviewed 75 studies and reports and included interviews with dental students, recent graduates, and academics. That work called for clear rules on acceptable AI use and structured teaching that prepares students to check outputs, recognise uncertainty, and remain accountable for their decisions. The emphasis on accountability is critical. AI can assist judgment, but it cannot absorb professional responsibility. In dentistry, as in the rest of healthcare, the clinician remains responsible for the decision taken in front of the patient.
Structured teaching would need to go beyond software familiarity. Students need to understand how AI systems are trained, where they can fail, and why a confident answer can still be wrong. They also need training in information appraisal: when to trust an output, when to verify it, and when to disregard it. That skill set is especially important in a field where visual interpretation and pattern recognition are central, because a system that appears accurate in straightforward cases may behave unpredictably when anatomy is complex, data quality is poor, or the clinical picture does not fit the pattern it has learned.
The survey findings suggest a curriculum problem rather than a student motivation problem. Demand exists. Curiosity exists. Confidence in AI’s future importance exists. The missing piece is formal instruction that keeps pace with the technology and makes clinical judgment harder to outsource. Training that stops at access and usability would leave a dangerous gap between familiarity and competence.
There is also a cultural dimension. Students who are introduced to AI only as a productivity tool may come to view it as a neutral authority. Students who are taught to question it will likely develop a healthier form of digital literacy: one that treats every output as provisional until it is tested against the clinical context. In a profession built on responsibility, that habit may matter as much as technical skill.
What the study does not show is whether current curricula are improving, or how different schools are handling AI teaching in practice. It does, however, show that the students already expect formal training and want clearer guidance on how to use AI safely and critically. The next question for dental education is not whether AI belongs in the curriculum. It is how quickly schools can build teaching that keeps pace with a tool set students are already treating as part of their professional future.