Why Healthcare AI Is 20 Years Behind
100+ carefully crafted slides, as a thank-you to my dedicated paid subscribers and amazing Founding Members.
Welcome to AI Health Uncut, a brutally honest newsletter on AI, innovation, and the state of the healthcare market. If you’d like to sign up to receive issues over email, you can do so here.
I’ve been fortunate to be invited to speak at quite a few healthcare AI events this year. Along the way, I’ve built 100+ carefully crafted slides. I’ve used parts of them to drive home some of my strongest points during recent talks, including the past two weeks at the Health AI Summit in Albuquerque, New Mexico and at Health2Tech in Vilnius, Lithuania.
Not a single soul has seen all of my slides yet.
Today I want to offer a small gift to my paid subscribers and especially to my Founding Members for their loyalty and support. I’m sharing all 100+ slides. (Well, 141 to be exact. 😉) Thank you for backing my mission to expose bad actors in health tech and healthcare AI, to dissect AI models and AI products in healthcare, and to ultimately make healthcare better for all of us.
But first, if you’ve been following my healthcare fraud investigations, including my three-part deep dive into Hippocratic AI — here, here, and here — you already know that much of my work comes from sources “behind the enemy lines.” These are the brave men and women inside these organizations who refuse to stay silent when they see things that are wrong, unethical, or outright illegal. They reach out to me because mainstream outlets either won’t challenge their own sponsors. Or hide behind the excuse of “not enough evidence.” Or simply don’t give a sh*t.
So I’m reaching out to all of you to help shine a light on two more organizations:
Commure (yet another shady Hemant Taneja’s creation)
Mayo Clinic (a massive institution where, yes, some things look good, but plenty does not)
I believe I already have enough to publish some deeply alarming pieces on both. Still, if you know any stories or insights, or can introduce me to people who do, about what’s happening inside these organizations — illegal, unethical, or, on the flip side, constructive and positive — please don’t stay silent.
If you want to share, everything remains strictly confidential. I’ve been doing these investigations for almost three years. I would never reveal my sources. Not to the FBI. Not to the DOJ. Not to anyone.
Also, if you are in NYC on Thursday, December 4, don’t be a stranger. Register and say hi at Health2Tech, where I’ll be moderating a panel on innovation in healthcare AI, real tech adoption in clinical workflows, and what healthcare providers actually expect from AI.
OK, back to my slides… If you joined any of my lectures this year, you know I promised to share my slides with you. Today, I’m making them available exclusively to my paid Substack subscribers and Founding Members through this post. I’m putting the slides behind a paywall to keep them out of reach of LLMs and search engines, especially given how AI companies have been shamelessly stealing copyrighted material.
These 100+ slides represent months and months of meticulous research. Enjoy.
And a reminder. If you can’t afford this article — maybe you’re a student or between jobs — just reach out. That’s exactly why I created the AI Health Uncut Founding Member Club. Thanks to generous support from the people in that exclusive group of (currently) 19, I’m able to offer access to anyone who needs it.
If you’d like to become a Founding Member of the AI Health Uncut community, you can join through this link. You’ll be making a real impact and helping me keep pushing for better outcomes in healthcare through AI, technology, policy, and beyond.
Here are the titles of all 100+ slides (the real number is 141 😉):
1. Title Slide: Paging Dr. Watson: Why Healthcare AI Is 20 Years Behind
2. Disclaimer: Cite My Work, Please!
3. Reach Out to Me on Social Media
4. WellAI, My Startup (Merged with Chart2Chart)
5. AI Health Uncut, My Substack Newsletter
6. My Dataset of 132 Digital Health Companies
7. I Critique Flawed AI Models in Healthcare…
8. …Especially Those Embarrassingly Hyped by Mass Media
9. I Investigate Healthcare Fraud
10. And of Course, I Expose Bad Actors in Digital Health
11. Selected List of Companies I’ve Investigated
12. I Also Cohost the “Digital Health Inside Out”
Podcast With Alex Koshykov13. I’m Also on the Advisory Board of PeriOptima.ai
14. I’m Also on the Board of Health2Tech, a Nonprofit That Organizes Health Tech Events Around the World
15. Health2Tech Branches Around the World
16. Google’s Transformers (2017) and Covid (2020) Had a Far Greater Impact on AI Adoption Outside of Healthcare
17. But Maybe Things Are Finally Turning Around for AI Adoption in Healthcare…
18. Even Though on the Ground It Sure Feels Like We’re Far Behind in Medicine
19. The Healthcare AI Paradox
20. Treating the Sick Is Profitable. Keeping People Healthy Isn’t.
21. Healthcare Innovation Is Being Killed As Problems Run Rampant
22. Why Does It Take Just a Few Lines of Code to Access the Most Advanced AI Model Ever Built,…
23. …While the Entire Healthtech Industry Has Spent Over a Decade Failing to Reliably Access Patient Records?
24. American Healthcare Is Complex — But It Doesn’t Have to Be
25. American Healthcare is Also Big. It is the #1 Employer and the #2 Contributor to GDP After Technology.
26. Two Primary Barriers to Healthcare Innovation
27. The Margin and the Mission Must Align
28. What Makes Innovating in Healthcare Hard?
29. Customers Make Innovating Hard
30. Incumbents Make Innovating Really Hard — The Market Is Highly Concentrated
31. Incumbents Make Innovating Really Hard — Their Size Is Massive
32. Incumbents Make Innovating Really Hard — Every Company Is A Healthcare Company These Days
33. Incumbents Make Innovating Really Hard — They Hate Innovation
34. Incumbents Are Swallowing the Healthcare System Piece by Piece
35. Epic Has No Real Incentive to Innovate
36. Regulators Make Innovating Hard — Regulators Capture Rules Healthcare
37. Regulators Make Innovating Hard — Policy, Not Innovation, Drives American Healthcare
38. EU’s AI Laws Are Even Tougher
39. History of Artificial Intelligence
40. Warner V. Slack, MD, Pioneer of Patient-Centered Computing & Medical Informatics
41. Eliza, the First Medical AI Agent
42. Why Machines Don’t Need to Think Like Humans
43. These 13 Influential Voices and 9 AI Companies Think AI Will Replace Doctors
44. While It’s Fashionable to Claim That AI is More Accurate Than Doctors,…
45. ...Most AI Accuracy Claims in Healthcare Have Been Debunked!
46. Let’s be honest, AI still ain’t ready for healthcare yet… A Story of Pruning Shears.
47. This Banana Was Diagnosed With Melanoma by 3 Out of 3 Leading AI Dermatology Image-identification Apps
48. LLMs Weren’t Built for Medical Diagnostics. Using Them as Such Could Be Dangerous.
49. The Cough That Broke 8 LLMs
50. AI Hasn’t Solved Healthcare’s Problems Because the System of Incentives Is Broken
51. Geoff Hinton’s 2016 Claim That AI Would Replace Radiologists “Within Five Years” Aged Poorly. There Are ~40,000 Radiologists in 2025, Up From ~34,500 in 2016.
52. Why Do AI Predictions for Healthcare So Often Fail? Because They’re Usually Made By People Who Are Nowhere Near Healthcare.
53. As With Any New Invention, Physicians Won’t Disappear. They’ll Learn AI.
54. AI Isn’t Replacing Clinicians. It’s Changing What “Being a Clinician” Means.
55. If Anyone Gets Replaced by AI, It’s the Useless Consultants 😄
56. Fast Forward to 2025: OpenEvidence Takes the Stage
57. OpenEvidence: Healthcare’s Most Data-Intensive Scientific Engine
58. Fast Forward to 2025: OpenAI Develops AI Consult
59. OpenAI’s AI Consult: The Man-Made Babysitter for LLMs
60. OpenAI’s AI Consult: A Human + AI Safety Net, Not a Standalone AI
61. Meredith Whittaker: Brutal Truth on LLMs, Tech Power & Privacy
62. Most Startups Fail
63. We All Know About High Startup Failure Rates
64. Not Failing Means You’re Not Working Hard Enough
65. You Miss 100% of the Shots You Don’t Take
66. Tech-enabled Healthcare Services Margins Falls Somewhere Between Traditional Healthcare and Traditional Software
67. Who Pays For Digital Health? In Theory, Multiple Parties…
68. In Reality, Most Startups Sell to Providers
69. What Is the Business Model of Digital Health?
70. The Importance of Aligning the Margin and the Mission
71. Startups Are Risky, Therefore VC Is Risky
72. Non-VC-Backed Startups Either Die Quickly or Eventually Beat VC-Backed Ones
73. VCs Have Caused Billions in Losses for U.S.-Listed Digital Health Companies Over the Past 5 Years
74. The Story Remains Consistent Over Time: VCs Continue to Pick Sh*tty Digital Health Companies Year After Year
75. Even If a Company Has Found a Sweet Spot, It May Not Want VC Funding
76. Out of the Top 10 Most Successful Publicly Traded Digital Health Companies, Only Two Are VC-Backed
77. Bankrupt Digital Health Companies Have Overwhelmingly Been Backed by VC and PE Firms
78. The Magic of Venture Capital in Digital Health: The Great Disappearance Act
79. VCs Are Playing Straight Into the Hands of Healthcare Monopolies by Eliminating Competition
80. Averages Don’t Tell Sh*t: While VCs Often Murder Digital Health Startups, Even Most Non-VC-Backed Startups Are Doomed to Die
81. VCs Can’t Pick Winners Anymore. Instead, They’re Picking Mediocre Startups They Plan to Unload Onto the Next Sucker.
82. VC Underperformed Equities and 60/40 in Every Possible Period Over the Past 25 Years!
83. VC and PE Returns Have Been Disastrous. And Nobody’s Even Mentioning Risk-Adjusted Returns—Unlike Stocks, You Can’t Just Hit a Button and Pull Your VC Money Out.
84. VC Distributions Are Hovering Near Their Lowest Levels in Decades
85. Morgan Stanley: VC Returns Plummet to 28-Year Low
86. If You’re an LP, You Now Have to Wait Over 10 Years Just to Get Your Money Back!
87. “Traditional” VC Playbook: The Infinite Loop of Fundraising
88. VC Funding Is Killing the Startup Ecosystem
89. “VC Pump & Dump”: Valuation Cycle of a Typical Digital Health Startup
90. Increasing Allocations to VC and PE at the Expense of Public Markets Has Been a Disastrous Move for Pension Funds
91. CalPERS Lost 24.8% on Their VC Positions in 2022. What Did They Do Next? They Didn’t Just Double Down. They Sextupled Down!
92. Since 1997, VC Investors Have Invested More Money Than They Received Back From VC Funds
93. “You Can’t Be Fired by Investing in General Catalyst”
94. When You Can’t Pick Winners if Your Life Depended on It, What Do You Do? You Build an Empire, Because at That Point, 2% Is Bigger Than 20%.
95. The Only Truthful Thing Hemant Taneja Has Ever Said
96. What Good Is VC Diversification Without Liquidation?
97. And of Course, VCs Also Pat Themselves on the Back by Sponsoring Fake Rankings
98. The Healthcare VC Mafia Org Chart
99. Digital Health’s Political Cartel Org Chart
100. The Healthcare VC Mafia Rules
101. Over Half of Late-Stage Deals Originate From VC Referrals. These Firms Leverage Their Elite Networks for Superior Deal Flow. And Yet, They Still Consistently Underperform in the Long Run.
102. How Venture Capital Is Cannibalizing the Economy
103. Digital Health Unicorns Have Disappeared—Because Mediocrity Trumps Innovation
104. But Maybe 2025 Is the Turning Point?
105. The IPO Party Is Over
106. The VC Culture of Mediocrity and the Incumbents’ Suppression of Innovation Have Created an Industry Dominated by “AI Tourists” and “AI Wrappers”
107. Is the Death of the AI Scribe Imminent?
108. Not a Single AI Scribe Has FDA Approval. It’s the Wild Wild West of Medicine.
109. I Haven’t Seen Any Real Innovations in Digital Health
110. Primary Care Is in the Worst Shape Ever—and the Lack of Digital Health Innovation & Shrinking Reimbursement Rates (Despite Healthcare Inflation!) Ain’t Helping
111. Sexism in Venture Capital: What Future Awaits My Daughters?
112. 21-Factor Quantitative Dashboard for Health AI IPO Valuation
113. 8 Digital Health Failure Patterns
114. Toxic Culture Is Eating Digital Health for Breakfast
115. Livongo: One of the Biggest Snake-Oil Deals in Financial Market History
116. Why Teladoc’s Acquisition of Livongo Was Disastrous
117. Theranos: One Drop of Lies, A Sea of Fraud
118. Better Theranos? People Sure Have Short Memories.
119. Hippocratic AI: $9/Hour “AI Nurses” and the Web of Lies
120. Babylon Health: The Madoff of Digital Health
121. Wall Street Analysts Were in Bed With Babylon’s Management
122. “Stock Markets Are Either Too Optimistic or Too Negative, but Eventually, They Settle on the Right Value.” Babylon’s Value Has Settled at $0 Four Months Later…
123. Olive AI: The Poster Child of “Champagne and Cocaine” Spending
124. October 2021 Was Peak Olive: “Frisson by Olive” Fragrance and the Notorious “Olive Bus”
125. Truepill: The Poster Child for VC Extortion
126. Truepill’s Trajectory: From Founders’ Fantasy to VC Nightmare
127. Truepill’s Founders’ Dilution via VC Extortion: An 8-Step Program From 100% Ownership to 0%
128. Suki: Digital Health’s Poster Child for “AI Tourism”
129. Suki: When There’s Zero Innovation, “Borrowing” Someone Else’s Tech Ain’t “Hard at All”
130. 7 Digital Health Success Patterns
131. Veeva Systems: The Secret to Success—Say No to VCs
132. Doximity: Wall Street’s Pariah, Doctors’ Darling
133. Startup Success Formula
134. My Mission and Exploring Possible Solutions
135. Staying Active on Policy Issues is Important
136. Solutions to U.S. Healthcare Crisis
137. Medical AI Success Depends Not on Raw Computational Power, But on Sophisticated Knowledge Curation and Conflict Resolution
138. Multimodal AI for Next-Generation Healthcare
139. A Rare Healthcare AI Success Story. Smartwatches That Actually Work for Atrial Fibrillation (AF) Detection.
140. Another Healthcare AI Success Story: Digital Twin GPT (DT-GPT) — LLMs Forecasting Patient Health Trajectories.
141. The Future of AI in Medicine
