Promptly Speaking AI, Misdiagnoses, and the Unsexy Work that Could Fix Healthcare with Matt Crowson, MD
In this episode of Promptly Speaking, Sara and Dan speak with Matt Crowson, a fellowship trained ear surgeon who has spent the last decade working at the intersection of clinical medicine, data, and artificial intelligence. Matt explains why diagnostic error is far more common than most patients realize, using ear infections as a powerful and personal case study. He shares how a machine learning model he co-developed at Mass General achieved roughly 95% accuracy in diagnosing pediatric ear infections, compared to the average human accuracy, which is closer to 60 to 70%, and why that gap matters for both everyday care and global health.
The conversation goes beyond diagnostics to focus on where AI is already making meaningful progress in healthcare today, particularly in reducing administrative burden through tools like ambient documentation and message triage. Together, they explore the barriers to AI adoption in healthcare, including privacy, liability, fragmented data systems, and regulation, as well as the growing role of clinicians in shaping AI products.
The episode closes with practical advice for patients on how generative AI can be used responsibly right now to better understand medical bills, insurance documents, and care options, while acknowledging the real tradeoffs around data privacy and trust.
đĄ Topics Covered:
- Why humans are surprisingly bad at diagnosing common conditions
- How AI can reduce diagnostic variability without replacing clinicians
- Pediatric ear infections as a global health problem
- AI in low-resource and rural healthcare settings
- Administrative burnout and âpajama timeâ for clinicians
- Ambient scribes and inbox triage as early AI wins
- Payers vs providers and how healthcare actually works
- Accountability and liability when AI is involved in care
- Why healthcare AI moves slower than other industries
- How patients can use generative AI today without over-trusting it
â± Timestamps:
00:00 Introduction to ear infections and AI in healthcare
00:44 Meet Matt Crowson: from ENT surgeon to AI advocate
01:36 Payers vs providers and how the healthcare system functions
02:39 AI and administrative burden in clinical work
04:55 Privacy, safety, and regulatory barriers to AI adoption
09:37 What a Chief Medical Officer does in an AI company
13:37 AI in rural healthcare and pediatric ear infection case study
26:35 Accountability and liability in AI-assisted care
26:51 The scalpel analogy and human responsibility
29:57 AIâs potential impact on healthcare costs
30:41 Generative AI and patient empowerment
36:10 What the future of AI in healthcare may realistically look like
36:40 How AI tools are implemented inside hospitals
38:35 Real-world AI deployments and auditing41:14 Vendor and healthcare system partnerships
42:01 Practical advice for patients using generative AI
45:07 Final reflections
How to Find Matt:
LinkedIn: https://www.linkedin.com/in/matthewgcrowson
Follow Sara & Dan:
Sara: https://www.linkedin.com/in/saralynneroberts/
Dan: https://www.linkedin.com/in/danroberts27/