Munjal Shah Advocates for AI “Super-Staffing” to Address Healthcare Worker Shortages

At the recent 2023 HLTH conference, Munjal Shah, CEO of startup Hippocratic AI, proposed a bold vision for utilizing artificial intelligence to alleviate worsening healthcare staffing shortfalls. Munjal Shah  argues that while diagnosis should remain solely the province of human clinicians, AI systems like large language models can take over many non-diagnostic tasks currently performed by nurses, dietitians, patient navigators, and other overburdened medical staff.

The annual HLTH event in Las Vegas brings together leading voices in healthcare innovation to discuss new technologies like AI. This year, the prevalence of generative AI systems able to convincingly converse like humans took center stage. During his talk, Shah contended that these conversational systems present a huge opportunity to provide patient-facing services at a fraction of the cost of human labor. He calls this concept “super-staffing”.

The idea is straightforward. According to the World Health Organization, healthcare is facing a global shortage of workers, projected to reach 10 million by 2030. Meanwhile, large language models can deliver certain routine services for as little as $1 per hour without risk of fatigue or burnout. For Shah, this makes AI an obvious solution for supplementing human capabilities in a resource-constrained system.

Shah believes AI should not replace clinicians‘ diagnostic judgment. But he sees abundant possibilities for AI to take over other time-consuming tasks: providing post-discharge instructions, explaining insurance benefits, delivering negative test results sensitively, answering routine patient questions, and more. “You can’t call every patient two days after they start every new medication. But at this cost structure, maybe you can,” Shah suggested.

On a panel titled “There’s No ‘AI’ in Team,” Shah and other experts agreed healthcare needs both human expertise and AI support. The key is combining both strengths. Humans are essential for oversight and specific skills, while AI offers unlimited scale and consistency at a low cost. This “centaur” approach allows AI to cover ground humans alone cannot.

For safe and effective AI, training methodology is critical. Hippocratic AI has hired thousands of clinicians to judge system responses and give reinforcement when mistakes occur. This human feedback teaches the AI to avoid hallucinating incorrect medical information. Extensive training on expert materials also instills proper reasoning. Shah stresses human collaboration at every step, from initial design through ongoing improvement.

Recent research indicates patients may even prefer AI-generated responses over human ones for qualities like empathy. But Shah maintains clinician review is vital, given potential risks in healthcare. The role of AI is to multiply access to support services, not provide fully autonomous diagnosis or treatment recommendations.

In Shah’s view, generative AI is ideal for conversational tasks and broad reasoning across information – precisely the skills required for patient-facing interactions. And given worsening staff shortages, virtual assistants able to provide personalized guidance at a massive scale are urgently needed. What Shah calls “super-staffing” could make comprehensive, high-quality care available to millions who lack it today.

At HLTH and beyond, Munjal Shah is working to steer AI in healthcare away from risky autonomous diagnosis and toward proven applications augmenting human capabilities. Thoughtfully designed AI, he argues, can fill staffing gaps, increase health equity, and bring expert-level support to all patients who need it.