AI is becoming ubiquitous, much faster than anyone anticipated, especially after the rise of large language models. In clinical care, radiology for example, AI has been a game changer. Clinicians are using AI for image analysis, to detect borders, to identify objects of interest and automate tasks that were previously manual, leading to faster and more accurate diagnoses. Some things are so nuanced that the human eyes alone cannot tackle. AI technologies help extract more information than ever. And it’s not displacing the humans, it’s actually helping them be more efficient, more more productive.

In research, the availability of massive datasets allows for new analytical methods, enabling researchers to identify patterns and trends that were previously impossible to detect. In education, AI supports a new kind of teaching called “precision education”. We can track student performance in real time to ensure comprehensive exposure to necessary content and better learning outcomes. AI allows us to see what they are exposed to in terms of content. What was the variety of content in their training? What exactly are they learning, and are they improving how they think and perfect their conclusions?

Generative AI, like ChatGPT, has made the benefits of AI more accessible and understandable to the broader public. These models learn from vast amounts of text data, mimicking human communication and generating text, which is particularly useful in clinical documentation and patient interactions. However, AI is much more than ChatGPT and CoPilot. AI encompasses other techniques like deep learning and machine learning now.

One major advancement is “ambient notes,” where a microphone records conversations between clinicians and patients. AI transcribes, analyzes, and extracts relevant codes or insights for diagnoses and therapies. This technology allows clinicians to focus more on patient interaction rather than typing notes, improving both patient and physician experience. Transitioning to this system requires adaptation and retraining for clinicians and students however. Clinicians need to think about how they document what they are doing differently.

One significant concern is the potential for AI to increase disparities between large, well-resourced healthcare systems and smaller organizations that simply can’t afford the technology investment. Other challenges include privacy, confidentiality, cybersecurity, and the complexity of integrating multiple technologies. However, the cost of cloud computing for AI, and this goes for research as well, is a rapidly growing expense. In the past, a system could buy some new servers, enhance memory of their current servers, etc. Those days are gone.

No, current training programs, including PhDs and fellowships, are not producing enough experts to meet demand. Beyond training more informaticians and digital leaders, it’s crucial to raise digital literacy across all roles in healthcare, as AI will impact everyone, not just clinicians. Jobs will change for everyone working in healthcare.

Priorities should include:

• Building the right infrastructure, especially for secure and affordable cloud computing.
• Updating curricula to ensure students are prepared for a future where AI is integral to clinical practice and research.
• Ensuring the sustainability of investments, as adopting AI will increase costs in already expensive medical training of our students.

There is often a tendency among technology leaders to focus on successes and present an overly optimistic view, while real challenges and issues may be downplayed or ignored. In reality, most institutions are facing significant difficulties in adapting to the rapid changes brought by AI.