/https://megaphone.imgix.net/podcasts/342bc2ea-4300-11ef-8b66-df65d463a1a7/image/b74d74edefe25379e16f1b805e76cda9.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Bendy Bodies with Dr. Linda Bluestein
Dr. Linda Bluestein
What AI Gets Wrong About Your Health with Michael Turken, MD (Ep 213.5)
SEP 15, 202666 MIN
/https://megaphone.imgix.net/podcasts/342bc2ea-4300-11ef-8b66-df65d463a1a7/image/b74d74edefe25379e16f1b805e76cda9.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Description
Patients are already asking AI the questions they cannot always ask their doctors.
That creates enormous opportunity, but also real risk. AI can help people organize complex medical information, prepare for appointments, and better understand their health. It can also sound completely certain when the evidence is weak or the answer is wrong.
In this live episode of Bendy Bodies, host Dr. Linda Bluestein speaks with Michael Turken, MD, an internist at UCSF and founder of My Doctor Friend, an AI health companion designed to support patients navigating medical complexity.
Dr. Turken explores the growing gap between what patients need from healthcare and what traditional systems can realistically provide, especially between appointments. He explains where AI may help fill that gap, where it can go wrong, and why these tools should support rather than replace the patient-physician relationship.
The conversation includes practical strategies for connecting medical records to AI platforms, checking what an AI model actually understands about your health history, keeping that information accurate and current, and recognizing situations where AI may be less reliable, particularly when the science itself is still evolving.
Takeaways:
AI chatbots are not healthcare providers, and many are not subject to HIPAA protections in the same way healthcare organizations and covered entities are.
AI can sound highly confident even when its answer is incomplete, overly reassuring, or wrong. The conversation includes discussion of AI sycophancy, in which a model may reinforce a user’s assumptions or provide reassuring responses rather than appropriately challenging them.
If you connect your medical records to an AI platform, periodically ask it to summarize what it believes about your diagnoses, medications, major medical events, and current health history. This can help uncover outdated, missing, or incorrect information.
AI may be particularly useful in addressing the gap between what patients want from healthcare, such as the ability to ask questions at any hour, organize complex information, and prepare for appointments, and what traditional healthcare systems can currently provide.
For medically complex patients, the quality of an AI answer depends heavily on the quality, completeness, and accuracy of the information the model is using.
To get 2 months of Premium access to My Doctor Friend, use code "BENDY"
Want to learn more about My Doctor Friend?
Website: https://about.mydoctorfriend.ai/
Instagram: @mydoctorfriend.ai
Want more Dr. Linda Bluestein, MD?
Website: https://www.hypermobilitymd.com/
YouTube: https://www.youtube.com/@bendybodiespodcast
Instagram: https://www.instagram.com/hypermobilitymd/
Facebook: https://www.facebook.com/BendyBodiesPodcast
X: https://twitter.com/BluesteinLinda
LinkedIn: https://www.linkedin.com/in/hypermobilitymd/
Newsletter: https://hypermobilitymd.substack.com/
Shop my Amazon store https://www.amazon.com/shop/hypermobilitymd
Dr. Bluestein's Recommended Herbs, Supplements and Care Necessities: https://us.fullscript.com/welcome/hypermobilitymd/store-start
Want to learn more about the UVA EDS Center?
For Appointments and Questions: [email protected]
UVA EDS: https://www.uvahealth.com/healthy-practice/advancing-care-through-ehlers-danlos-clinic
UVA EDS FAQ: https://www.uvahealth.com/support/eds/faq
UVA Pediatric Integrative Medicine: https://childrens.uvahealth.com/specialties/integrative-health
Thank YOU so much for tuning in. We hope you found this episode informative, inspiring, useful, validating, and enjoyable. Join us on the next episode for YOUR time to level up your knowledge about hypermobility disorders and the people who have them.
Join YOUR Bendy Bodies community at https://www\.bendybodiespodcast\.com/.
YOUR bendy body is our highest priority!
Learn more about Human Content at http://www\.human-content\.com
Podcast Advertising/Business Inquiries: [email protected]
Part of the Human Content Podcast Network
FTC: This video is not sponsored. Links are commissionable, meaning I may earn commission from purchases made through links
Chapter:
00:00 AI as Malpractice
00:28 Welcome and Guest Intro
01:20 Boutique Sponsor Break
01:54 Why Build My Doctor Friend
05:52 Choosing AI Platforms
07:15 Record Linking Pitfalls
09:21 Always Ask What It Knows
12:35 Three Models for Safety
15:50 Privacy Reality Check
19:29 Reassurance Trap Lawsuit
21:47 Making One Model a Jerk
23:27 Urgent Care MRI Story
27:18 Fixing Sycophancy Prompts
32:12 Doctors React to AI
35:26 Avoiding AI Misinformation
38:51 Privacy and Medical Records
41:46 Finding EDS Savvy Doctors
46:24 Talking AI With Your Doctor
48:46 My Doctor Friend Demo
55:31 AI in Medicine Next 5 Years
58:41 Carbon Footprint and Wrap Up
Learn more about your ad choices. Visit megaphone.fm/adchoices

