AI in Medicine Doesn’t Replace the Doctor – It Makes the Patient the Decision-Maker
The public debate about AI in medicine gets stuck on a false choice: either the doctor decides, or the algorithm does.
But that’s not how good medicine has ever worked. The best clinical decisions have always balanced three things: evidence and guidelines, clinical experience, and what the patient actually values. AI doesn’t change that—it just makes that balance visible for the first time.
The Dyad Problem
Traditionally, a medical consultation is a dyad: doctor ↔ patient.
This structure is inherently asymmetrical. The doctor holds knowledge, training, and authority. The patient holds symptoms, fear, and the actual consequences of the decision. The patient is alone on one side of an expertise gap they can’t fully cross.
This creates two failure modes:
Paternalism: “Trust me” becomes the default answer when the doctor runs out of time or patience to explain trade-offs.
Passivity: The patient nods, doesn’t ask the hard questions, and later regrets the choice because they didn’t understand their own values in the decision.
The Third Voice
Now introduce a third party—not an autonomous “AI doctor,” but a transparent, auditable reasoning partner.
Suddenly, the patient is no longer alone opposite expertise. The patient becomes the observer of a dialogue between clinical judgment and an evidence-based system that can show sources, assumptions, uncertainty, and alternatives.
The patient moves from object to subject. From recipient of decisions to participant in them.
This is what we call triadic medicine: not doctor-and-algorithm, but doctor-and-evidence-system-and-patient, where reasoning becomes visible and negotiable instead of hidden inside someone’s head.
Why Dentistry Is the Perfect Test Case
Dentistry—especially implantology and surgical dentistry—lives in one of the most “decision-dense” zones of medicine. A single consultation involves:
- Data richness: CBCT scans, 3D imaging, anatomical mapping
- Risk sensitivity: nerve proximity, sinus floor, bone volume, complications
- Preference dependence: time vs. invasiveness, cost vs. longevity, aesthetics vs. surgical complexity
With 3D implant navigation (developed by Dr. Michael Truppe), the anatomy is already translated into a precise plan. The next step is making the reasoning behind that plan—and the alternative paths—equally transparent.
A typical implant consultation with triadic medicine looks like this:
Patient: “I want fixed teeth. I’m scared of surgery. And I don’t want surprises.”
Clinician: “Let’s map your anatomy, discuss options, and plan safely. There are multiple good paths—but they differ in risk, timeline, and trade-offs.”
AI (Virtual Patient AI with MedlibreGPT): “Based on your imaging and factors, here are the option sets:
- Option 1: Single implants with guided 3D navigation—higher precision, staged timeline
- Option 2: Augmentation + implants—longer treatment, potentially improved foundation
- Option 3: Alternative prosthetic strategy—shorter timeline, different long-term trade-offs
Here are the risk factors to discuss explicitly (e.g., nerve proximity, sinus floor, bone density). Here are typical complication rates from the literature. Here are the key patient preference questions that change the recommendation.”
Now the patient asks better questions:
- “How close is the implant to the nerve in my case?”
- “What does ‘risk’ mean numerically—and what’s the downside if it happens?”
- “If the AI and doctor disagree, what assumption is causing that?”
- “Which path best fits my values: fewer visits, lower invasiveness, or maximum longevity?”
This is empowerment: not patients making medical decisions alone, but patients seeing the reasoning.
The Freud Frame: Why Externalizing Evidence Works
Freud described three forces in the psyche:
- Id: desire, fear, urgency—what the patient feels
- Ego: reality-testing and practical judgment—what the clinician does
- Superego: rules, norms, professional standards—what should be done
In traditional medicine, clinicians carry all three. They’re the empathic listener (id), the practical decision-maker (ego), and the internal guideline engine (superego).
That’s a lot to carry—and it’s exactly where biases, shortcuts, and paternalism happen. The clinician’s own preferences leak into the superego.
With transparent AI, the superego becomes external. Evidence and norms are visible, discussable, and negotiable instead of hidden inside the clinician’s reasoning. The clinician is freed to do what only humans do well: context, nuance, trust, and values.
The patient no longer guesses what the professional standards actually say. They see them.
The Virtual Patient AI Model: Precision + Explanation
Virtual Patient AI isn’t “AI dentistry.” It’s precision dentistry with transparent decision-making.
3D implant navigation is the clinical anchor—it turns anatomy into a plan, and a plan into execution with documented precision.
The triadic consultation completes that story: you don’t just perform precise treatment, you make the why and the trade-offs visible and patient-readable.
Trust becomes more than reputation. It becomes auditability.
The “Third Man” Risk: What We Deliberately Refuse to Build
In Carol Reed’s 1949 film The Third Man, Harry Lime (Orson Welles) is an enigmatic, mostly invisible figure who corrupts every relationship simply by existing in the shadows. His presence introduces hidden agendas, disrupts trust, and transforms straightforward human connections into something sinister and opaque.
AI in medicine risks becoming exactly that kind of “third man” unless we deliberately design it otherwise.
The corrupting version: An AI that operates as a black box, speaks directly to patients without physician oversight, prioritizes corporate profit over patient health, or forces physicians to follow recommendations they know are wrong. In this scenario, the patient faces not one authority (the physician) but two, with one unknowable.
The enabling version: An AI that is transparent in its reasoning, cites its evidence, allows itself to be questioned by both physician and patient, remains subordinate to human judgment, and genuinely serves shared decision-making. In this scenario, the third man is a visible presence—a trusted consultant in the room—who contributes specialized knowledge without claiming authority.
The boundary condition is clear: In triadic medicine, AI must never become a black box, a profit-optimizing hidden actor, a system that bypasses the clinician, or an authority that overrides patient values.
Triadic medicine only works if:
- The AI is explainable and auditable
- The clinician can challenge it
- The patient can ask “why?” and get a real answer
What Changes for Patients
Patients don’t care about frameworks. They care about what changes for them:
Clarity: Fewer vague statements. More “here are the options and trade-offs.”
Confidence: Trust based on transparent reasoning, not blind faith.
Control: Values are explicitly included. “What matters most to you?”
Safety: Disagreements become visible quality checks.
Continuity: Documented reasoning supports follow-ups and second opinions.
This is especially powerful in dentistry because many decisions are genuinely preference-sensitive. The “best” implant path isn’t universal—it depends on what you value.
The Real Question
The wrong question is: “Will AI replace the doctor?”
The right question is: “Will we use AI to make reasoning visible—so patients can participate as equals?”
That’s the Virtual Patient AI stance:
- Dr. Truppe delivers precision care and human clinical judgment
- Virtual Patient AI externalizes evidence and structure
- The patient becomes the central decision-maker, supported—not abandoned
The future of medicine isn’t doctor-vs.-algorithm. It’s doctor-and-patient-as-partners-with-evidence-made-visible.
That’s triadic medicine. And it starts with refusing to hide the reasoning.
Why On-Premises AI Is the Key to Making This Real
Vienna, December 22, 2025
Dr. Michael Truppe
Why On-Premises AI Is the Key to Making This Real
The Virtual Patient AI model needs an on-premises AI supercomputer because triadic medicine only works when all three actors—clinician, patient, and AI—operate inside the same trusted environment. If AI reasoning runs in a distant cloud, the core promise of transparency breaks: sensitive CBCT data, photos, and preference profiles leave the practice; vendors and third parties sit invisibly in the loop; and neither doctor nor patient can fully oversee what the system is doing with their information.
With on-premises deployment, all clinical data stays inside the practice, under the direct control of the treating dentist, enabling strong privacy, regulatory compliance, and a clear audit trail of every recommendation.
It is referenced to the patient database and the skill level of the dentist. Just as importantly, local inference makes the reasoning immediately available chairside: the AI can explain in real time why it ranks one implant strategy over another, which anatomical risks it is weighing, and how different patient priorities (time, invasiveness, aesthetics, cost) change the proposed plan.
The clinician can challenge those assumptions, the patient can literally see the logic on screen, and no hidden actor can override medical judgment or patient values. In other words, the on-premises Virtual Patient AI is not just a technical choice—it is the infrastructural condition that keeps the “third voice” visible, question-able, and firmly subordinate to human care.
Literature
PS: Claus Cristian Carbon schlug die Idee vor, die trilogische Beziehung zwischen Patient, Arzt und KI für das „Virtual Patient AI“-System zu evaluieren.
Univ.-Prof. Dr. Claus-Christian Carbon
Universität Bamberg
Fakultät für Humanwissenschaften und Erziehungswissenschaften
Lehrstuhl für Allgemeine Psychologie und Methodologie
Wir arbeiten gemeinsam an der Pilotstudie „Psychomorphometrie und faziale Ästhetik“, einer prospektiven Studie, die von der Ethikkommission der Medizinischen Universität Wien unter der EK-Nr. 1749/2016 genehmigt wurde und welche 2026 nun mit KI erweitert wird.
Discover more from Ordination Dr. Michael Truppe
Subscribe to get the latest posts sent to your email.