Unexpectedly low accuracy of GPT-4 in identifying common liver diseases from CT scan images

Tracing the trajectory of artificial intelligence (AI) in medicine reveals a tapestry of innovation, with threads reaching back to the late 20th century when rudimentary computer programs first offered diagnostic assistance[1]. However, it wasn't until the emergence of deep learning that the true narrative of AI began to unfold—a tale where its capabilities to support diagnosis and decision-making started to be vividly imagined and explored.