a16z co-founder Marc Andreessen said in a New York Post interview that "Doctor ChatGPT" is already a better doctor than 99% of physicians, fueling debate over AI's role in medicine.
June 2026 · AI in Medicine
"Doctor ChatGPT" Already Beats 99% of Doctors, Says a16z's Marc Andreessen
The investor's categorical claim — backed by his self-driving analogy that AI need only beat the median human — reignites debate. Benchmark scores are strikingly high, but clinicians insist the tool remains an aid, not a replacement.
USMLE Accuracy — GPT-4o vs Humans
Dashed reference: USMLE passing line ≈ 60%. GPT-4-class models clear every step by a wide margin.
79%
of the time, professionals rated ChatGPT's patient answers higher quality & more empathetic (JAMA, 2023)
80–95%
GPT-4-class score range across USMLE steps — well above the ~60% pass line
99%
of doctors that Andreessen claims AI already outperforms — a contested figure
Where it helps
Available any hour of the night
Answers from multiple angles
Handy for an initial assessment
Useful second-opinion / diagnostic aid
Why it's not a doctor
Misdiagnosis & over-reliance risk
No physical examination possible
No legal liability
Struggles with tabular data & complex pathophysiology
"Judge AI not by perfection, but by whether it beats the median human."
The emerging consensus: powerful as a diagnostic aid, but not a replacement. Serious symptoms still belong with a physician — clinical use of AI remains largely supplementary.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…