Andreessen: Is it true that Doctor ChatGPT outperforms 99% of doctors?

      Marc Andreessen has made a bold assertion regarding health care. The billionaire investor claims that “Doctor ChatGPT” is already superior to 99% of human doctors. This statement faces strong opposition from medical professionals and lacks substantial peer-reviewed support.

      Andreessen's comment was made during an episode of Joe Rogan’s podcast in early June. The New York Post highlighted the statement, which quickly gained traction after the prediction-market account Polymarket shared it on X on June 29. The assertion is striking, yet the evidence supporting it is minimal.

      The assertion itself lacks data backing the 99% claim. As the head of Andreessen Horowitz, a leading venture capital firm in technology, his words carry significant influence. He stated to Rogan, “99% of the time, the answer that I’m getting from the AI is better than I would get from talking to basically almost any expert.” The Post reformulated this into the assertion about doctors.

      Many are indeed utilizing this technology. OpenAI reports that over 40 million individuals consult ChatGPT about their health on a daily basis. Given that scale, even a small error rate could lead to millions of potentially dangerous responses, especially for individuals unable to afford actual medical consultations.

      However, research presents a more nuanced view. A 2024 study published in the Journal of Medical Internet Research had both doctors and ChatGPT-4 respond to 100 actual patient inquiries. While patients found the chatbot to be more empathetic and useful, specialists identified 15 of its responses as potentially harmful, according to IBTimes UK, cautioning that lay readers might struggle to differentiate between safe and dangerous answers.

      A more recent study raises further concerns. A February 2026 paper in Nature Medicine, led by Mount Sinai urologist Dr. Ashwin Ramaswamy, evaluated OpenAI’s ChatGPT Health using 60 clinical scenarios. The tool failed to direct 51.6% of real emergencies to the emergency department, instead suggesting a routine appointment. One missed instance involved a patient experiencing respiratory failure.

      This aligns with a known trend. AI has demonstrated the ability to match or surpass doctors in identifying likely diagnoses but falls short in evaluating risk and determining treatment options, which are critical components of real medical practice.

      Financial interests also play a role in Andreessen's assertions. His firm has invested in health-AI startups, such as Hippocratic AI, Ambience Healthcare, and Abridge, and manages a $500 million biotech fund supported by pharmaceutical giant Eli Lilly.

      This doesn’t invalidate his claims, but it indicates that his optimism may not be impartial. His financial benefit from belief in Doctor ChatGPT warrants a more careful examination of his statement.

      Additionally, there are reasons to question his understanding of the tool. In May, Andreessen posted an extensive “super prompt” on X instructing ChatGPT to “never hallucinate or make anything up” and to verify every fact. Engineers highlighted a significant flaw: instructing the AI to avoid hallucination cannot actually eliminate that tendency, as it's inherent to the functioning of large language models. AI expert Gary Marcus noted this as both amusing and somewhat alarming, given that he had not grasped this concept.

      The implications of this discussion are significant. Millions of individuals are treating a fluent, self-assured chatbot as if its articulateness equates to medical accuracy, which is not the case.

      Doctors do not claim that AI is ineffective. Many are currently using it, and diagnostic tools are continuously improving. Dr. Adam Rodman, a Harvard hospitalist researching AI in medicine, offers a sensible guideline: avoid using a chatbot for emergency triage and consider it as a supplementary tool to a human consultation, rather than a substitute.

      This is the gap in Andreessen’s soundbite. Public skepticism towards AI exaggeration is already prevalent, and caution regarding its use in health matters is justified. While Doctor ChatGPT may indeed have its benefits, it currently does not outperform 99% of physicians based on existing evidence.

      This article discusses medical choices and personal health. While AI tools can serve as a beneficial starting point, for any urgent or serious health issues, it’s crucial to consult a qualified healthcare professional.