Nvidia and AI healthcare company Hippocratic AI have announced a partnership to develop generative AI-powered healthcare agents designed to assist with non-diagnostic tasks. These AI agents, powered by Nvidia’s technology and trained on Hippocratic’s specialized healthcare-focused language model, are designed to enhance patient interactions through fast, empathetic responses. Tested by over 1,000 registered nurses and 100 physicians in the U.S., the AI agents outperformed not only leading AI models like OpenAI’s GPT-4 and Meta’s LLaMA 2 70B but also human nurses in several key areas. For instance, they demonstrated higher accuracy in identifying medication impacts on lab values (79% vs. 63%), detecting toxic drug dosages (81% vs. 57%), and recognizing disallowed over-the-counter medications for specific conditions (88% vs. 45%). Nvidia and Hippocratic AI highlight the technology as a potential solution to the ongoing healthcare workforce shortage, emphasizing its ability to improve efficiency while maintaining human-like interactions.
Hippocratic AI, co-founded by Munjal Shah, is positioning its AI agents as a cost-effective alternative to human nurses, with an operating cost of just $9 per hour compared to the U.S. median nursing wage of $39.05 per hour. Shah stated that the collaboration with Nvidia aims to refine the technology and improve healthcare access and equity. Nvidia’s vice president of healthcare, Kimberly Powell, reinforced the vision of AI-driven voice agents transforming patient interactions by mimicking human responses. While the AI-powered agents are not designed for diagnostics, they offer potential benefits for administrative and support roles in healthcare, providing a scalable solution to staffing challenges. However, the deployment of AI in critical healthcare settings raises questions about patient trust, regulation, and the long-term implications of replacing human labor with AI-driven alternatives.