CT Lin, MD, professor of internal medicine, at the University of Colorado Anschutz School of Medicine, and a secondary faculty member in the Department of Biomedical Informatics, explored these themes during the closing keynote at the 2026 American Medical Informatics Association (AMIA) Amplify Conference in Denver, CO, on May 21, examining how generative AI is reshaping clinical care, operational workflows and the broader human experience of medicine.
Generative AI in healthcare is often discussed as either revolutionary or dangerous, but that binary misses the complexity of the moment. “I’ve started thinking about generative AI in healthcare in three categories: the mundane, the profane and the insane,” said Lin. “Together they capture the full emotional and practical spectrum of how humans are encountering AI today, from ‘I haven’t tried it yet’ to ‘I’m vibe-coding my entire life.’”
Healthcare workers are no exception. Some are cautiously experimenting with simple prompts, while others are redesigning workflows and building tools faster than institutions can evaluate them.
The mundane category may actually be the most important. These are the practical, accessible use cases that help colleagues get started and immediately see value. Clinicians are using generative AI to summarize meetings, draft patient education materials, reduce inbox fatigue and assist with documentation. Operational leaders are exploring ways to automate repetitive administrative tasks that contribute to burnout.
“These applications may sound small, but they matter because they reduce friction in systems already stretched beyond capacity,” Lin explained. “The mundane is where trust begins."
The profane category is more unsettling. These are the stories and research findings that challenge long-help assumptions about human primacy and professional oversight. Stanford researchers recently reported that AI alone sometimes outperformed humans using AI, raising uncomfortable questions about the “human in the loop” model that healthcare has long treated as a gold standard.
“If humans supervising AI does not always improve outcomes, what exactly does that mean for clinical decision-making and expertise?” Lin said. “Are humans supervising machines, or are machines beginning to supervise humans?”
At the same time, generative AI introduces familiar but amplified risks, including hallucinations, automation bias and overreliance on systems that sound authoritative even when they are wrong. Healthcare organizations are now confronting ethical and operational questions that do not yet have clear answers.
Then there is the insane category – the glimpses of a future that feels implausible until is suddenly becomes real. One recent example involved an Australian technologist who reportedly designed, built and administered an experimental cancer vaccine for his dying dog using AI-assisted methods.
“Read that sentence again,” Lin added. “Not a pharmaceutical company. Not a university lab. A motivated individual empowered by generative AI.”
Regardless of how that particular story is ultimately judged, it points toward something larger: generative AI is dramatically compressing the barriers to experimentation and innovation. Capabilities once limited to elite institutions are becoming available to motivated individuals with internet access, software tools an curiosity.
“Some of these efforts will fail. Some will prove exaggerated,” Lin explained. “But some may fundamentally reshape how discovery happens in medicine and biotechnology.”
The challenge for healthcare leaders is that all three categories are unfolding simultaneously. Health systems are being asked to adopt expensive AI assistants while also creating entirely new frameworks for validation, benchmarking and governance. Traditional evidence-generation models move slowly, while generative AI systems evolve almost weekly.
“We are not simply evaluating another software platform,” Lin said. “We are building a new worldview for how intelligence, expertise and care may function together in the future.”
Clinical environments demand safety and reliability, yet innovation often requires experimentation and uncertainty. The questions emerging around AI in healthcare are bigger than productivity or automation. They touch economics, ethics, professional identity and even what it means to practice medicine in an era of increasingly capable machines.
“This is a whole new world,” Lin shared. “We are having to build a new world view, a new set of benchmarks and a whole new financial model to consider adopting expensive AI assistants in clinical care.”
Lin’s closing keynote at the 2026 AMIA Amplify Conference will invite attendees to think together about the mundane, the profane and the insane future of generative AI in healthcare.
And yes, there will also be a new ukulele parody song about AI in healthcare.