Tuesday, April 16, 2024

Artificial Intelligence in Healthcare: A Journey through Three Epochs

Similar articles

Artificial intelligence (AI) is currently at an unprecedented peak of interest, attracting attention from academia, media, and the general public. In the healthcare sector, leaders face the challenge of deciding when and how to deploy AI, as well as understanding its potential benefits and associated risks. A recent study on this subject sheds light on potential benefits and associated risks.

The capabilities and risks of different AI types vary significantly. To avoid misleading healthcare decision-makers, a simple framework divides AI into three distinct epochs, helping them grasp the strengths, weaknesses, and challenges of each.

Symbolic Artificial Intelligence and Probabilistic Models

For over 50 years, AI primarily focused on encoding human knowledge into rules and decision trees. While achieving notable successes, such as IBM’s Deep Blue defeating a chess world champion, this symbolic AI had limitations, including human logic errors and bias. It struggled with real-world complexities and was eventually complemented by probabilistic models.

The Era of Deep Learning

In the early 2010s, deep learning powered by neural networks revolutionized AI. It enabled computers to ‘see’ images, translate languages, and more. In healthcare, it excelled at identifying diseases from images and predicting outcomes from electronic health records. However, these models learned from labeled examples and could falter with real-time data mismatches and biases.

Artificial Intelligence in Healthcare

Foundation Models and Generative Artificial Intelligence

AI 3.0, marked by models like large language models, transcends task-specificity. It adapts to diverse tasks without retraining and exhibits enhanced capabilities, including complex question interpretation, content generation, and human-like responses. These models impact daily life in various applications, including healthcare, where they answer medical questions and assist with documentation.

Foundation models and generative AI represent a significant AI evolution, offering the potential to transform healthcare. Decision-makers can benefit from understanding these three AI epochs, each characterized by distinct capabilities and risks.

 

Original Article DOI: 10.1001/jama.2023.25057

Original Title: Three Epochs of Artificial Intelligence in Health Care

Subscribe to our newsletter

To be updated with all the latest news, offers and special announcements.

Latest article