At the recent HLTH 2023 conference, Munjal Shah, CEO of Hippocratic AI, spoke about how artificial intelligence could help address healthcare staffing troubles. Shah sees generative AI as a way to care for more patients, calling this concept “super-staffing.”

The annual HLTH event in Las Vegas brings together leaders in healthcare and technology to discuss innovations in the field. This year, much of the focus was on the new possibilities that advances in generative AI opened up. During a panel titled “There’s No ‘AI’ in Team,” Shah argued that while diagnostic applications of AI remain risky, there are significant staffing shortages in non-clinical roles where AI could make an impact.

According to Shah, a worldwide deficit of healthcare workers is only expected to grow in the coming years. At the same time, comprehensive and high-quality care requires more than just diagnosis – it also relies on services like chronic care nursing, scheduling, diet planning, and patient education. AI that can provide some of these non-diagnostic services safely and affordably could help bridge staffing gaps.

The panelists agreed that AI alone cannot solve all healthcare problems. But used judiciously with human oversight, AI could make a difference in manageable areas of need. For Shah and Hippocratic AI, staffing shortages represent one area ripe for an AI augmentation approach.

To train generative AI responsibly for healthcare, Hippocratic AI has brought on thousands of medical professionals to test systems and provide human feedback. This “centaur” approach combines the pattern recognition of AI with expert human judgment. With enough quality training data from medical experts, AI can learn to generate helpful and empathetic responses like a human would.

Shah refers to this concept of exponentially expanding virtual healthcare staffing through AI as “super-staffing.” While a human nurse may cost $100 per hour, an AI assistant may only cost $1 per hour and can interact with unlimited patients simultaneously. This could allow more patients to receive individualized services like post-op care, medication counseling, and chronic condition monitoring that are impossible to scale with human staffing alone.

Generative AI like ChatGPT has advanced conversational abilities that can replicate the human touch critical in healthcare interactions. A recent study even found that participants preferred ChatGPT’s responses to patient questions over those written by doctors, based on measures of quality and empathy.

As Munjal Shah explained, generative AI’s strengths lie in its conversational capabilities and ability to reason across diverse documents – ideal traits for patient-facing applications. While specialized AI exceeds singular tasks, generative models can hold more human-like dialogues and conclude from varied sources like a knowledgeable human.

The goal is to partially replace human healthcare workers but augment them with AI assistants that can help cover more ground. AI proxies working with human oversight could manage tasks that may not be feasible for overburdened staff to provide for every patient.

In Shah’s view, generative AI has opened up the possibility of giving every chronic disease patient a dedicated nurse for medication management and lifestyle coaching. Or they provide in-depth genetic counseling to all needy patients instead of just a few. For situations where conversing like a caring human is paramount, AI may finally offer a path to providing high-touch care to all.