Unraveling How Generative AI and Agentic AI Can Reshape the Scope of Digital Health: The Good, The Bad, and The Ugly

Abstract

Artificial intelligence is revolutionizing the field of healthcare at an ever-increasing rate; yet, the fundamental question does not merely revolve around whether artificial intelligence can carry out clinical or administrative functions. Rather, it is concerned with how generative artificial intelligence and agentic artificial intelligence are redefining the entire realm of digital health. This shift brings along major considerations, as although there exists great potential in such innovations to make care more efficient, adaptable, and inclusive, they could also bring about considerable challenges with regard to accuracy, safety, privacy, biases, and accountability. Hence, both the strengths and weaknesses of such technologies merit examination in the context of real-world digital health applications. To achieve this goal, this research employs a qualitative approach in carrying out a case study analysis of nine specific examples, namely documentation assistance, early warning systems, psychiatric electronic health record interpretation, medical translation, radiological reporting optimization, imaging processes, algorithmic discrimination, privacy concerns, and inappropriate reactions from therapy chatbots. This paper will then proceed to consider these case studies by means of Bentham’s felicific calculus to illustrate the pros and cons of digital health solutions. It appears that while generative artificial intelligence and agentic artificial intelligence can lead to improved workflow, greater accessibility, and regular care, their true worth lies in their application that ensures safety, justice, privacy, and trustworthiness.