AI misdiagnoses over 80% of early medical cases, study finds

A new study tested AI chatbots' ability to diagnose patients in a clinical setting and found that the technology wasn't quite up to the task yet, the Financial Times reported.

In a tweet, the Financial Times (@FT) indicated that misdiagnoses occurred "in over 80% of early medical cases" evaluated during the research.

AI chatbots misdiagnose in over 80% of early medical cases, study finds https://t.co/PfbPwqDcxd

— Financial Times (@FT) April 13, 2026

The study, published in JAMA Network Open on April 13, tested 21 chatbots using "29 standardized clinical vignettes," its abstract explained.

Among the chatbots tested were cutting-edge models, including OpenAI's GPT-5, Anthropic's Claude 4.5 Opus, Google's Gemini 3.0 (Flash and Pro), and Grok 4.

While uploading images improved models' accuracy, the authors found that the AI chatbots struggled with the step-by-step process of identifying a condition from the information provided.

In particular, researchers observed AI chatbot failure rates of over 80% for differential diagnoses, when a diagnosis is complicated by symptoms matching more than one underlying illness, injury, or condition.

Ultimately, the authors reported improved chatbot performance in later diagnostic stages, with misdiagnosis rates falling below 40%. The most capable models demonstrated accuracy rates over 90% for final diagnoses.

Beyond the study's findings, the general public's faith in AI and related technologies had dwindled significantly at the time of publication, for reasons ranging from job market disruptions to growing controversy over data centers.

Initially, data centers were viewed as a regional nuisance, but by the end of 2025, their impact on utility bills had become a key issue nationwide. Electric bills skyrocketed, with monthly costs exceeding housing payments in some areas, as PBS reported.

On X, commenters were unsurprised by the study's findings.

Which of these savings plans for rooftop solar panels would be most appealing for you? Save $1,000 this year 💸 Save less this year but $20k in 10 years 💰 Save less in 10 years but $80k in 20 years 🤑 Couldn't pay me to go solar 😒 Click your choice to see results and earn rewards to spend on home upgrades.

"Keep this in mind the next time some AI CEO spews nonsense like AI will replace everyone," one replied.

"Glorified autocomplete apps should not be diagnosing medical cases," another said.

"Human insight remains the gold standard," a third wrote.

Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.