Healthcare educators must prioritize critical thinking as AI becomes standard in medical practice
Medical students need to understand generative AI well enough to decide when to use it, but not so well that they outsource their own thinking. That balance defines the central challenge for health professions educators as AI tools become embedded in clinical work.
The evidence on AI's impact in education remains mixed. Some studies show it enhances learning; others suggest it degrades students' ability to think independently. Rather than choosing sides, educators should focus on building AI literacy-a foundation that lets students make informed decisions about which tasks warrant AI assistance and which don't.
What AI literacy looks like in healthcare education
Medical students should know what generative AI is and how it works. They need to understand its strengths, limitations, ethical implications, and how to evaluate its outputs for accuracy.
They should also grasp the material costs of AI infrastructure. Cloud systems supporting large language models consume significant energy and water resources. Understanding this allows students to weigh whether specific uses justify the environmental trade-off.
Beyond generic AI competencies, healthcare students need context-specific knowledge. They should recognize which types of AI are used in clinical settings, spot opportunities to apply AI responsibly, and ask developers the right questions about real healthcare problems.
Most critically, students must develop the confidence to question AI outputs and override them when clinical judgment demands it. This requires strong foundational skills in evidence-based reasoning, comfort with ambiguity, and metacognitive awareness-the ability to reflect on their own thinking processes.
Five strategies to embed critical thinking
1. Teach reasoning explicitly. Make your thinking visible. Have students generate and test hypotheses before turning to an LLM. Focus on mechanisms and causal understanding, not just answers.
2. Distinguish between types of AI use. Not all AI assistance undermines thinking. Some tasks benefit from outsourcing; others don't. Teach students to reflect on when they're offloading cognition unnecessarily.
3. Use LLMs as teaching tools. Incorporate them into case-based learning to explore bias, hallucination, and flawed reasoning. This builds critical interrogation skills.
4. Adopt a "think first" approach. Encourage students to use LLMs for feedback and alternative perspectives, not to complete work. Design assessments that reward reasoning and explanation over right answers.
5. Build metacognitive habits. Have students use structured reflection. Can they explain their work without AI? What did they learn? Are they improving over time?
The human in the loop must be prepared
Regulatory frameworks for AI in healthcare rely on humans to oversee AI decisions. But a human in the loop only works if that person has developed the knowledge and critical capacity to catch problems.
Health professions educators bear responsibility for ensuring that happens. The foundational skills are not novel-critical thinking, data literacy, and self-awareness are teaching staples. What's new is the deliberate focus on preserving these skills in an environment where outsourcing to AI is always an option.
Learn more about AI for Education and AI for Healthcare to explore how these principles apply across different sectors.
Your membership also unlocks:
