On the one-year anniversary of the public launch of ChatGPT, the AI in health care’s technological evolution is poised for a groundbreaking shift. In a commentary published in JAMA on November 30, 2023, Dr. Robert Wachter, Chair of UC San Francisco’s School of Medicine, and Dr. Erik Brynjolfsson, Director of the Digital Economy Lab at Stanford University, delve into the transformative potential of generative artificial intelligence (genAI) tools. This story explores their insights, shedding light on how genAI could revolutionize healthcare, overcoming longstanding challenges faced by the industry.
Accelerating healthcare’s digital revolution with GenAI
The healthcare sector has been notorious for its slow adoption of transformative technologies. From the hesitant integration of electronic health records (EHR) to the unsuccessful attempts with IBM’s Watson Health, the industry has faced hurdles in embracing change. Yet, the authors argue that genAI, capable of producing high-quality content distinct from its training data, possesses unique properties that might expedite the transformation process.
Dr. Wachter, known for chronicling health information technology challenges, believes genAI can break the “productivity paradox” cycle that has hindered the adoption of general-purpose technologies in various industries, including healthcare. The historical resistance to such changes, driven by misaligned incentives, complexity, privacy regulations, and a general aversion to change, is now met with a potential solution in genAI.
The productivity paradox of information technology, coined by Dr. Brynjolfsson in 1993, highlighted the delayed gains in productivity despite the widespread adoption of general-purpose technologies. In the context of healthcare, genAI is seen as a solution due to its user-friendly nature, minimal hardware requirements, and alignment with existing digital workflows.
Unlike the challenges faced during EHR adoption, genAI’s ease of use positions it favorably in an environment where healthcare professionals and patients are already engaged with digital tools. Also, the current readiness of the healthcare ecosystem, accustomed to using digital data and systems, creates an opportune moment for genAI to address clinical and business needs effectively.
AI in health care – applications and challenges in AI integration
The historical failures of AI applications in healthcare, particularly in the 1960s to 1980s, were attributed to attempting to replace the doctor’s cognitive functions. GenAI’s early applications, but, focus on alleviating administrative burdens, such as scheduling appointments, medication refills, and answering patient queries. For healthcare professionals, genAI is expected to assist in generating clinical notes, prior authorization requests, and summarizing complex patient records.
While there is potential for genAI to aid in diagnosis, the emphasis is on suggesting possible diagnoses rather than replacing the expertise of physicians, acknowledging the high stakes and consequences associated with healthcare decisions.
Despite its promise, genAI faces challenges that need addressing for successful integration into healthcare systems. The technology itself must continue to improve, especially as it expands into high-stakes clinical realms. The integration of AI into EHR systems, while more accessible than before, still requires refinement. Also, the cost implications of adopting AI in healthcare need to be managed effectively to ensure a return on investment.
Labor-management tensions around AI, exemplified by recent strikes in other industries, could pose challenges, but the critical shortages and burnout levels in healthcare might mitigate some pushback. As AI ventures into clinical territories, finding a balance where healthcare professionals collaborate effectively with technology becomes paramount to success.
Paving the way for GenAI success
For genAI to succeed in healthcare, regulatory frameworks must be established, particularly for high-stakes clinical applications. Yet, the complexity of regulating general-purpose technologies poses a daunting challenge. Differentiating between regulating a specific AI algorithm and overseeing AI’s broader role in the entire care system requires innovative approaches to ensure patient safety and quality of care.
As genAI holds the potential to reshape healthcare, the industry must navigate these challenges collaboratively. How can the healthcare ecosystem strike the right balance between embracing the transformative power of genAI and ensuring responsible and effective integration into clinical workflows?
