NVIDIA have developed an AI nurse that outperforms living humans

NVIDIA, in close collaboration with Hippocratic AI, has created an artificial intelligence that can perform the functions of a nurse better than real people. The development is aimed at solving the global “staffing shortage” in the healthcare industry.

NVIDIA’s medical AI agents are based on the Polaris neural network. It provides a smooth and natural conversation with patients in a friendly tone. The main advantages of the solution are low latency and confident recognition of natural speech.

Hippocratic CEO Munjal Shah is confident that generative AI will help address the most pressing needs of modern medicine. In particular, it can be used to solve problems such as staff shortages and the provision of affordable and quality medical care. Kimberly Powell, NVIDIA’s senior executive overseeing healthcare, believes that all of this is only true if an important condition is met – the technology must respond to patients the way a human would. A recent test of Polaris showed that it outperformed ChatGPT-4, LLaMA-2 70B and even real-life U.S. nurses and therapists on key conversation parameters.

Against the backdrop of a growing healthcare labor shortage in the US (which could reach 10 million by 2030), the development seems more than timely. However, there is a view that neurosthetics will never be able to fully replace nurses in hospitals, no matter how advanced they are. Nurses do much more than what is prescribed in their job descriptions. Even though an hour of neural network work will cost less than $9 and an hour of nurse work more than $42, hospitals will not give up using the labor of living people. But AI nurses could become really useful assistants, doing some of the work of human staff.

Read also: A new Polaris is about to erupt in the sky