Multimodal generative AI advances precision health by continuously learning from diverse data streams, highlighting the crucial role of clinicians and human expertise in shaping future healthcare AI models.

The rise of ChatGPT has sparked immense interest in generative artificial intelligence, especially in healthcare. There's ongoing discussion around foundation models and multimodal generative AI, leading to some confusion. The first part of this series is focused on foundation models.

A small number of medical AI tools dominate usage, highlighting the pressing need for equitable access and improved outcomes in AI-driven healthcare

Advancements such as federated learning and generative AI have implications for broader precision medicine practices.

Three key lessons for regulating disruptive medical technologies, though AI's rapid evolution indicates a need for novel regulatory paradigms, possibly including self-regulation.

There is an urgent need for innovation in clinical research, and whilst AI has huge potential to offer improvements, there is a necessity for a robust regulatory framework to ensure reliability and equity.

The rapid growth of large language models holds promise for assisting clinical decision-making, but they cannot, and should not, take place of all the combined cognitive and perceptive capabilities of clinicians.

While AI holds vast potential for healthcare, slow adoption stems from data hurdles, workflow disruption, and a lack of long-term vision.

Oppenheimer's atomic bomb and AI share duality, lack of regulation, and uncertainty, but AI can become a force for good, revolutionizing diagnosis, treatment, and patient care.