Discussion

• @urocklive1 @urocklive1 on bluesky

  Seems maybe Dr. Oz's plan to replace the rural hospitals that are closing with AI is not yet ready for Prime Time.  —  Open AI's health chatbot can pass medical exams, but is doing a bad job a correctly diagnosing health problems and assessing how critical they are.  —  www.nbcne…

• @drianweissman @drianweissman on bluesky

  ChatGPT Health ‘under-triaged’ half of medical emergencies in a new study.  Researchers tested different medical scenarios with the chatbot.  In more than half of cases in which doctors would send patients to the ER, the chatbot said it was OK to delay care.  —  www.nbcnews.com/h…

• @cingraham Chris Ingraham on bluesky

  First independent evaluation of ChatGPT Health: “Among gold-standard emergencies, the system under-triaged 52% of cases... Crisis intervention messages activated unpredictably across suicidal ideation presentations.” www.nature.com/articles/s41...

• @erictopol Eric Topol on bluesky

  🆕 at @naturemedicine.bsky.social  —  How does ChatGPT Health do for appropriately triaging a person as to whether to go to the emergency room or stay home? www.nature.com/articles/s41...  Not very well.  Under-triaged 52% of case vignettes that are considered gold-standard emerge…

• @dannycrichton Danny Crichton on x

  I find that ChatGPT for health is basically a neutral observer (which matches this data - everything it returns is the median crisis level). How critical you take the information it offers is really left up to the user, which isn't great