At the recent HLTH 2023 conference, Munjal Shah, CEO of Hippocratic AI, advocated using artificial intelligence to help alleviate worsening healthcare staffing troubles. He believes generative AI could enable “understaffing” to provide services like chronic care nursing at scale.

The annual HLTH event in Las Vegas brings together leaders across healthcare to discuss innovations. This year’s primary focus was leveraging rapid AI advances, specifically large language models (LLMs), to transform care delivery. Shah highlighted staffing deficits as a prime opportunity during the “There’s No ‘AI’ in Team” panel.

The shortage is severe, with the WHO forecasting a global healthcare worker shortfall of 10 million by 2030. Shah argues generative AI can help fill gaps by working alongside human providers in a “centaur” model combining the strengths of both. Hippocratic AI aims to deploy LLMs for non-diagnostic tasks like patient education, care coordination, and counseling.

Shah contends LLMs are not yet reliable for diagnosis, as they can “hallucinate” incorrect or dangerous health information. However, they excel at conversing naturally and reasoning across documents. These capabilities make them well-suited for non-clinical roles.

The panelists agreed AI alone cannot fix healthcare but has potential in targeted applications with human oversight. Munjal Shah believes virtual “understaffing” could dramatically expand access to overlooked areas like chronic care management.

Human feedback is critical to training safe, trustworthy healthcare LLMs. Hippocratic AI has thousands of clinicians judge system responses to ensure they mirror expert quality and empathy. Over time, reinforcement learning helps the LLM converse appropriately.

By augmenting professionals with AI that costs just $1 per hour and never burns out, Shah believes previously unthinkable levels of high-quality care can be provided. For instance, patients with multiple chronic conditions could have a dedicated nurse for medication management.

Other applications include explaining bills, providing genetic counseling, delivering test results, and answering surgery-related questions. The goal is not to replace staff but to enable them to reach more patients through AI supplementation.

Recent research found that participants favored ChatGPT’s patient education responses over physicians’ in quality and empathy.  Munjal Shah says conversational ability and cross-document reasoning make LLMs well-suited to patient interactions.

Munjal Shah advocates thoughtfully applying generative AI to help healthcare successfully navigate mounting staff shortages. He believes LLMs working collaboratively with professionals can enable “understaffing” and dramatically expand access to services often out of reach today. The future Shah envisions is where top-quality, empathetic care can be provided to all who need it through human-AI teamwork.