How AI is redefining the role of critical thinking in medical education

Artificial intelligence has quickly become a foundational part of the clinical toolkit, especially when it comes to the more tedious aspects of documentation and other administrative requirements. The ability to transcribe conversations in real-time, retrieve information from the depths of the medical record, and draft clinical content almost instantaneously has revolutionized the way physicians interact with digital data. 

The industry has breezed past the question of whether physicians will use AI to augment their clinical workflows. It’s starting to generate convincing evidence that doing so can produce better results on metrics that matter, including provider burnout and patient experiences.   

But it must take a moment to pause at the next big hurdles: how can health systems, professional societies and medical schools train the first generation of AI-native physicians to use these tools responsibly? What role will critical thinking and clinical analysis play in an environment where some AI is pushing up against the highest levels of accuracy…while other tools are still lagging dramatically behind? Ultimately, what will it mean to be a physician when an AI companion is always available to take on some of the workload?  

At Meharry Medical College, a historically black medical school in Nashville, clinical educators are zeroing in on these challenges as they design and deliver medical education that recognizes and embraces the fact that AI can be an incredibly valuable partner for clinical care – as long as physicians understand how to use it responsibly. 

“You cannot separate a physician from data anymore,” said Fortune Mhlanga, PhD, Founding Dean of the School of Applied Computational Sciences and Professor of Computer Science and Data Science. “That means you also cannot separate a physician from the expertise in data analysis and critical thinking that is required to leverage digital tools in their daily workflows.”  

“At Meharry, we see data science as something deeply transformational that should be embedded across research, education, and clinical care, from the time a student enters medical school to the time they retire from practice. Every physician must have these competencies, because every physician is going to be interacting with AI throughout their entire career from this point on.” 

Teaching physicians to interrogate data and ask critical questions about AI outputs must be an essential component of medical education, agreed Sajid Hussain, PhD, Chair and Professor of the Department of Computer Science and Data Science in Meharry’s School of Applied Computational Sciences. 

“It’s simply not optional any longer,” he stressed. “Physicians need to be able to ask where a recommendation comes from. Is it explainable? Is it biased? Does it generalize to my patients?” 

“In the past, physicians relied on published literature and trusted that peer review had vetted the evidence. Now, even when AI systems claim to be explainable, we know that the methods used to measure explainability can introduce bias themselves. It’s a moving target, which means clinicians can’t simply accept AI recommendations at face value. That makes human oversight more important now than ever before.” 

Making the best use of the time AI can save

Widely adopted AI tools, such as ambient scribes, have been shown to reduce the time physicians are spending on administrative tasks and repetitive work. But what should physicians do with those extra minutes? 

“That time needs to be invested in higher-value cognitive work and critical analysis, including reviewing what AI tools are presenting to the physician,” stated Hussain. “AI may generate clinical notes that sound eloquent and impressive, for example, but clinicians have to read them carefully and challenge what they’re seeing. There is too much evidence that ambient listening may introduce errors and risks into the workflow.” 

“It’s important to make sure that we’re actually saving time while producing something valuable and worthwhile, instead of simply offloading your thinking to something that cannot be fully trusted on its own.” 

AI education should create power users, not programmers

AI-focused medical education isn’t about turning physicians into software engineers, Mhlanga and Hussain pointed out. Instead, the next iteration of medical school curriculum should aim on creating power users who fully understand how to apply AI effectively within their area of practice. 

“We have spent a great deal of effort to build workshops, modular training, and coursework that can be integrated into existing curricula rather than taking students away from their primary field of study,” said Hussain. “It’s really about adding to the education they receive, not replacing it with programming classes.” 

Meharry’s mission to model the optimal blend of clinical education and AI competency is even more notable in light of its historical focus on training black medical students, who are significantly underrepresented in the United States, said Mhlanga, and even more scarce in the ranks of data scientists developing and training medical tools. 

“One of the challenges we face is that fewer than 3% of people from underrepresented groups hold graduate degrees in fields like computer science, data science or AI,” he said. “As an institution focused on health equity, we’re working to close that gap.”  

“At the same time, we’re also thinking about how underrepresentation influences algorithmic bias. Many predictive models are trained on datasets that don’t adequately represent all populations. By preparing a more diverse generation of data scientists and AI professionals, we’re also helping ensure that future algorithms better serve everyone.” 

For institutions like Meharry, preparing physicians of all backgrounds for an AI-enabled future means teaching them more than just how to integrate AI into the workflow.  

As medical schools evolve to meet the demands of the AI-powered environment,  

truly impactful education must center on developing the critical thinking skills that help determine whether AI-generated output is accurate, relevant, and appropriate enough for patient care.

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].