You might think that critical thinking is a general-purpose skill that can be installed once and applied to any situation. In this view, you teach students “how to think,” they practice on a few exercises, and they walk away able to reason well across biology, psychology, machine learning, and artificial intelligence. The problem is that critical thinking is domain-bound, and so it’s imperative that the healthcare education system ground the critical evaluation of AI outputs within the domain of artificial intelligence. Asking medical students to evaluate the credibility of AI-generated responses for accuracy, bias, and relevance seems circular if they don’t know enough about the subject to have expectations in the first place.
In a case-based learning (CBL) environments, students learn by analyzing and discussing realistic scenarios, or ”cases,” that mirror real-world solutions. Rather than receiving information only through lectures, learners apply knowledge to solve problems, make decisions, and reflect on outcomes. Medical students are taught how to use analytical thinking to understand structure, relationships, and patterns of a phenomenon. And our education system does a pretty good job teaching systems analysis, but since the advent of generative AI, we have not delivered a corresponding shift in AI fluency that prioritizes critical thinking.
Students are taught to ask ”what’s the evidence?” But without domain knowledge, they often can’t tell good evidence from bad. By asking our students to think about the credibility of AI-generated responses, we’ll have introduced a more reliable tool in the doctor’s black bag.
The instructional implication, which surprises people, is that well-intentioned ”critical thinking courses” or genericworksheets of logical fallacies, taught without rich content, tend to produce weak transfer. A more effective approach is to interleave critical thinking and the main subject material, and show how AI reasoning is used there. This means teaching students to argue like doctors while practicing medicine and to assess evidence like scientists while conducting experiments. In other words, our future physicians and nurses would be better served if we embed critical thinking inside substantive disciplinary content.
The new information layer
Back when the internet and search engines came onto the scene, professors taught students web literacy skills and how to track claims back to their origins and assess the trustworthiness of those sources. Now that artificial intelligence changes how information is accessed, we should respond again with another shift in literacy education, and our medical schools are in a crucial position to guide our students through the new information layer.
How can leaders successfully facilitate transformative change?
Successfully facilitating transformative change rests on four operational pillars: the skills we prioritize, the people who carry the message, the evaluation habits we install, and the cognitive level we design toward.
Prioritizing core-skills
It begins with the skills themselves. We need to treat analytical thinking and critical thinking as equal priorities. The World Economic Forum’s Future of Jobs Report 2025 names analytical thinking as the most sought-after core skill, yet critical thinking never makes its core-skills list. That omission is why we can’t afford to ignore the importance of critical thinking skills if AI chatbots continue to confabulate outputs with sycophantic confidence.
People who carry the message
From there, the question becomes who builds trust in the technology, and the answer is the educators themselves. Those who teach are best positioned to model what responsible adoption looks like. As mentors, our educators can aid in identifying the tool’s capabilities and limitations, while weighing its impact on stakeholders, and challenging both their own assumptions and the developer’s framing. But here is where we should take a Socratic approach so that we empower our students to think independently.
The halo effect
Which brings us to the third pillar: protecting evaluation from bias. The halo effect, where a good overall impression influences how we view individual claims, can subtly weaken our ability to analyze arguments and lead us to incorrect conclusions. To ensure we’re making sound judgments, it’s important to guard against this effect. Subject-matter-experts would develop clear strategies for assessing AI-generated outputs, rather than allowing cognitive biases to rule our decision-making.
Higher-order thinking
Finally, all of this has to be anchored somewhere measurable. Institutions would target the upper three tiers of Bloom’s taxonomy: analysis, evaluation, and creation because those cognitive levels are precisely where critical thinking lives. When our learning objectives are built at that level, the skill we claim to value becomes the skill we actually develop. And the alignment of objectives to outcomes are directly visible.
Decision fatigue
An underlying concern is that evaluative work gives you decision fatigue. To mitigate this effect, we would convert decisions into rules. Every rule eliminates future decisions. If we are consistently evaluating AI responses for accuracy, bias, and relevance, then a rule for that would be, “Approve a response if it is factually correct, materially unbiased, and directly answers the user’s question. Otherwise, revise or reject it.”
Future-ready
While AI can accelerate work for faculty and staff, critical thinking remains essential for questioning assumptions, assessing outputs, and making sound decisions. When leaders adopt new technologies, it’s important to prioritize human judgment and create learning and change initiatives that enhance independent thinking, rather than just focusing on increasing tool usage.
By combining critical thinking with Large Language Models (LLMs), we can revolutionize healthcare by empowering physicians to assess complex situations, adapt treatment plans, monitor health risks, incorporate new information, streamline administrative tasks, accelerate scientific discoveries, enhance clinical decision-making, and provide personalized care in unprecedented ways. The convergence of artificial intelligence advancements and healthcare education will undoubtedly enhance the quality of life for individuals globally.