SAN FRANCISCO (KGO) — A recent study has unveiled that artificial intelligence systems can surpass human doctors in intricate medical reasoning tasks. This revelation opens up new inquiries regarding how clinicians, hospitals, and patients might utilize this technology moving forward.
Released on Thursday, the research delves into the performance of advanced AI “reasoning models,” evaluating their capabilities not just on traditional test-style questions but also on complex diagnostic scenarios that emulate real-world medical decision-making.
Dr. Jonathan Chan, a physician at Stanford and one of the study’s contributors, remarked on the remarkable pace of advancements in AI technology.
“The takeaway is that these models are becoming increasingly proficient at an astonishingly rapid rate,” Chan stated.
RELATED: Can artificial intelligence chatbots outperform human doctors? Here’s what new Stanford study found
Chan pondered, “This raises critical questions about where uniquely human attributes come into play and the extent to which we can rely on computers. The distinctions are becoming less clear.”
He clarified that the study is not aimed at providing a definitive answer regarding the necessity of human doctors. Instead, it underscores the challenges AI faces in replicating complex medical reasoning, particularly when cases extend beyond straightforward multiple-choice formats.
In detailing the intricacies of medical practice, Chan explained that doctors typically approach patients exhibiting various symptoms—such as fever and weight loss—by proposing potential explanations, assessing probabilities, and articulating their thought processes.
He acknowledged that AI systems have become increasingly proficient at navigating these complicated reasoning scenarios.
Nevertheless, Chan refuted the notion of entirely removing human involvement from clinical decision-making.
“Humans cannot be eliminated from the equation. Trust is a loaded term,” he remarked. “However, it would be unwise for me, or anyone else, to not utilize these tools to assist in diagnostics and medical reasoning. It’s akin to practicing without the internet.”
He drew a parallel between AI tools and online medical resources currently employed by doctors, suggesting that they can serve as valuable checks on conclusions, even if they are not infallible.
“We should definitely be leveraging AI to verify our findings,” Chan noted. “Not because AI is always correct, but humans are not infallible either. The synergy could lead to a more effective approach where we maintain responsible oversight.”
Chan indicated that there are specific areas where AI could increasingly assume responsibilities with minimal human oversight, particularly in administrative capacities.
He pointed to tasks such as documentation and summarization of hospital notes, emphasizing that physicians would be better utilized focusing on patient care rather than paperwork.
Notably, he mentioned that automated systems are already managing medication refills in states like Utah, a development that would have seemed far-fetched just a few years back.
Simultaneously, Chan expressed awareness of concerns circulating within the healthcare sector, highlighting apprehensions voiced by mental health professionals and the recent measures taken in California and Illinois regarding AI’s role in therapeutic settings.
“I completely understand the fear and anxiety surrounding this issue,” he stated. “It’s daunting. We must impose rigorous standards on these technologies because the stakes are high.”
The implications of the study also extend to patients’ interactions with technology in daily life, such as utilizing chatbots for clarification of medical summaries or terminology following brief consultations with doctors.
“It’s perfectly reasonable to inquire about your health,” Chan remarked, noting that AI can offer explanations for information that doctors may not have the time to elaborate on in detail.
Addressing concerns about potential job displacement due to AI, Chan suggested that while certain administrative and support roles may evolve, he doesn’t foresee a decrease in the need for clinicians.
“The complexity and responsibility of roles such as doctors, nurses, pharmacists, and mental health counselors are significantly greater,” he explained.
“While those roles will undoubtedly transform, the demand for human professionals remains abundant and essential.”
Moving forward, Chan highlighted the challenge of learning to leverage the expanding capabilities of AI while effectively managing the associated risks in healthcare.
Copyright © 2026 KGO-TV. All Rights Reserved.
