Podcast Summary

Solutions with Henry Blodget

Episode Title: Could AI Doctors Be Better Than Humans? This Physician Thinks It’s Possible.
Guest: Dr. Robert Wachter (Professor and Chair of the Department of Medicine at UCSF)
Date: March 2, 2026

Episode Overview

Henry Blodget sits down with Dr. Robert Wachter, an influential physician, author, and AI-in-healthcare optimist. Together, they explore whether artificial intelligence could improve health outcomes, expand access, and possibly surpass human doctors in some aspects of care. The episode delves into the history, current state, and future prospects of AI's integration into medical practice, balancing optimism with a healthy dose of reality about challenges and pitfalls.

Key Discussion Points & Insights

1. Dr. Wachter’s AI Optimism in Healthcare (02:53)

• Credentials & Perspective:
  Dr. Wachter clarifies he's not a blanket tech optimist—he once wrote a “grumpy” book about digital healthcare—but AI represents a unique opportunity due to:

  • Its rapid progress & applicability to healthcare’s “massive brokenness”
  • Immediate recognition (post-2022) that large language models could address real, daily problems.

• Quote:
  “AI is already widely used in healthcare and it will be increasingly so in the years ahead. And that’s good news, not bad.” (01:25, Wachter paraphrased by Henry)

2. Digital Transformation: History & Lessons Learned (04:53)

• Early AI Attempts:

  • 1970s/80s “if/then” rule systems failed due to medicine’s complexity.
  • Tackling diagnosis first was a mistake—should’ve started with simpler problems to win trust.

• Electronic Health Record (EHR) Revolution:

  • Digitization brought new issues: endless data entry, bureaucratic oversight, and "pajama time" (after-hours electronic work by clinicians) (09:31).
  • EHR patient portals gave access to more info but no context, generating anxiety and overwhelming doctors with emails.

• Quote:
  “We became very expensive, very grumpy data entry clerks.” (08:32, Wachter)

3. The Rise of AI “Digital Twins” & New Models of Care (11:40)

• Digital Doctor Examples:

  • Mayo Clinic is piloting “digital twins” of physicians—virtual assistants that interact with patients.
  • While full digital twins are a bit futuristic, the essential tech (AI-driven responses and triage) is already spreading.

• Access Transformation:

  • AI enables reaching patients far beyond a doctor’s physical location—historically impossible.

• Risks:

  • Deepfakes and misinformation: For every trustworthy digital twin, there could be a malicious or misleading one.

• Quote:
  “For every digital twin the Mayo Clinic produces that you know is going to be reliable... there’s another digital twin that could have me right now saying to you, ‘You should never get vaccinated. They will kill you.’” (12:10, Wachter)

4. The Psychology of Tech Adoption & the “Waymo Problem” (14:14)

• Public Skepticism:
  Change is met with fear—people highlight AI incidents over overall benefit. Analogy to driverless cars applies to medical AI, where inevitable errors shouldn't overshadow vast net positives.

• Quote:
  “Don’t compare me to the Almighty, compare me to the alternative...In many cases, the alternative is you can’t find a doctor or the doctor’s not very good.” (18:05, Wachter quoting Joe Biden)

5. Consulting AI as a Patient: Promise & Peril (19:11)

• Safety:

  • For simple queries, ChatGPT is “probably better than nothing.”
  • Pitfall: Patients lack expertise both in what to ask and how to interpret AI’s answers—context and nuance matter.
  • Future: Tools will know the patient’s history and ask follow-up questions, but they’re not fully ready today.

• Quote:
  “I think the tools for patients are not ready for primetime... Patients use these tools, and if they’re using the same tool, we’re asking them to do those two things, which they can’t generally do: understand what the key facts are... then interpret results, some of which may be great and some of which may be completely wrong.” (20:52, Wachter)

6. AI and Health Equity (23:20)

• Access Concerns:

  • Natural fear: Will low-income patients be relegated to seeing “just” the AI while the wealthy get real doctors?
  • Wachter’s take: AI actually stands to improve access for underserved communities, expanding quality care to millions who otherwise get none.
  • Analogy: TurboTax vs. live-accountants; both have value, and often “AI” is preferred for basic tasks.

• Quote:
  “I think over time... the ability to provide better access to people at a lower cost—I would bet on this (AI) over the alternative, which is trying to figure out ways of getting doctors into poor places... This has a greater possibility.” (24:15, Wachter)

7. What’s Working Now: AI Scribes & Summarization (28:31)

• Clinical Tools:

  • AI scribes transcribe and format visit notes—already in widespread, effective use (28:41).
  • Tools summarize massive, unmanageable patient records (some exceeding 700 pages) before encounters.
  • Automated drafting of discharge summaries for long-stay or complex patients.

• Risk:

  • “Human in the loop” oversight is crucial; over-reliance or “de-skilling” is a real concern.

• Quote:
  “The AI now is good enough to be very useful, but not perfect enough to be completely trusted.” (30:45, Wachter)

8. Training the Next Generation: Novice to Expert (31:53)

• Medical Education:

  • Worry over bypassing the hard-won learning needed to become an expert.
  • AI can aid teaching—monitor simulations, give feedback—but foundational knowledge & reasoning remain essential.
  • Risk: If we stop training doctors deeply, we may be left with AI-dependent novices.

• Quote:
  “There’s something different about the way experts use AI knowledge tools...If we bypass that stage because they don’t need to know that stuff anymore, they’re still novices...” (34:41, Wachter)

9. Resistance, Empathy & Human Touch (38:02, 47:39)

• AI Adoption:
  Young doctors generally welcome usable AI tools; older docs may be skeptical, but rapid adoption is driven by clear benefits.

• Empathy Surprises:
  Studies show AI sometimes beats doctors at empathetic response, mainly because:

  • It responds at length and with focus, not rushed by time or billing constraints.

• Quote:
  “The thing that bothered some doctors about those empathy studies is that the computer, the AI, has unlimited time.” (47:47, Wachter)

  “I don’t think empathy is going to be the great dividing line anymore.” (48:46, Wachter)

10. AI in Diagnosis: Finally Ready? (39:23)

• Current Use:
  Doctors use AI for diagnostic suggestions and to check for missed conditions—not to substitute but to augment cognition.

  • Integration with EHRs and real-time decision support systems is coming soon.

• Challenges:
  Context is crucial: Sifting out important vs. irrelevant information is still tricky for AI.

11. Five-Year Outlook (52:56)

• Predictions:

  • Routine care increasingly delivered or supported by AI (e.g., routine medication management, vaccine scheduling).
  • Patient-facing AI that’s contextual—knows your history, interprets results personally, asks smart questions.
  • Wearables and continuous monitoring data integrated into care.
  • In surgery and complex care, AI assists, but humans remain in the loop due to legal and trust needs.
  • Caution: As “easy” cases shift to AI, what’s left for doctors grows more complex; this must be anticipated.
  • Biggest worry: Deepfakes and misinformation. Trustworthy sources and systems will be essential.

• Quote:
  “In five years...certain problems will be managed primarily by AI... Patient portals will be much more robust and much more trustworthy, and you’ll be able to get care and advice tailored for you.” (53:16, Wachter)

Notable Quotes & Memorable Moments

• On AI’s Promise
  “This is really about how do we deliver the best health and best healthcare... Most conveniently and at the lowest cost to the most people.” (48:28, Wachter)

• On Future Doctor-AI Partnership
  “It’s not human or AI, one or the other, it’s binary...it is coming together.” (51:42, Henry)

• On Deepfakes
  “The thing I worry about the most is deepfakes. The idea that anybody can be made to look like they’re saying anything. And how can patients figure out what to trust and what not to trust?” (55:31, Wachter)

Key Timestamps for Important Segments

| Timestamp | Segment | |-------------|------------------------------------------------------| | 02:53–04:15 | Dr. Wachter’s AI optimism/context | | 04:53–08:51 | History of digital healthcare & early AI failures | | 09:31–11:20 | EHR burdens & unintended consequences | | 11:40–14:14 | Digital twins and access transformation | | 15:32–18:46 | Tech adoption psychology, “Waymo problem”, trust | | 19:11–23:20 | Patients using AI, safety and interpretation limits | | 23:54–24:34 | AI’s impact on health equity and access | | 28:41–31:05 | AI scribes, record summarizers—what’s working now | | 31:53–36:36 | AI and medical education: dangers & opportunities | | 39:23–44:35 | Diagnosis, clinical decision support, patterning | | 47:39–51:42 | Empathy: AI vs human, triage, empathy study details | | 52:56–56:15 | Five-year predictions, ongoing challenges | | 55:31–56:15 | Deepfakes & trust |

Tone and Style

The conversation is candid, thoughtful, witty, and practical. Dr. Wachter brings a blend of measured optimism and realism, emphasizing both the game-changing potential and the necessity for caution and careful oversight in AI’s rapid adoption in healthcare.

Conclusion

If you haven’t listened:
This episode offers a nuanced, real-world guide to how AI is already reshaping medicine—from addressing paperwork overload and supporting diagnosis, to extending care to the underserved, and even surprising findings about AI “empathy.” Dr. Wachter’s insights balance hope for better, more accessible care with grounded warnings about complexity, training, and digital trust. The future won’t be AI or humans—it will be both, and getting the mix right is the key to delivering better health for all.