Reflections from Thessaloniki, the city that shaped the first physician-philosopher
I am here in Thessaloniki, less than an hour's drive from Stagira, the birthplace of Aristotle, it feels fitting to ask a question that might initially seem absurd: what would the father of Western philosophy make of artificial intelligence transforming medicine?
Not absurd at all, as it turns out. Aristotle was not merely a philosopher. He was a biologist, an anatomist, and the son of Nicomachus, court physician to the King of Macedon. Medicine ran in his blood. He dissected animals, catalogued species, theorised on the human soul, and wrote more about living organisms than almost any other subject. If any ancient mind could engage seriously with the age of algorithms and neural networks reshaping healthcare, it would be his.
And I believe he would have both celebrated and cautioned us, with characteristic precision.
He would likely have admired the ingenuity. But he may also have challenged one of our deepest assumptions: whether intelligence alone is enough to make something truly wise.
Artificial intelligence has become the defining technology of our age. It writes essays, diagnoses disease, generates art, predicts protein structures, drives cars, and increasingly shapes human decisions. Yet more than 2,300 years ago, Aristotle posed questions that may still define the limits of machine intelligence today.
I have been invited to give a keynote on AI and Healthcare at the Aristotle Innovation Forum and to launch my book in Thessaloniki, in the epicentre of philosophy. As I look out into the Thermiac Gulf from the vantage point of The Makedonia Palace, I consider AI and its ramifications for our humanity; my thoughts have turned to this crucial topic.
The Physician Who Reasons, and the Machine That Calculates
Aristotle drew a crucial distinction between three forms of knowledge: episteme (scientific knowledge of universals and causes), techne (craft knowledge, how to make or do things), and phronesis (practical wisdom, knowing what to do in particular circumstances, and why it matters).
Modern AI in healthcare is extraordinary at episteme and techne. It can identify a malignant nodule on a CT scan with greater accuracy than most radiologists. It can predict sepsis hours before a clinician would recognise the signs. It can synthesise a patient's genomic profile against millions of data points to recommend a personalised cancer therapy. These are remarkable feats of pattern recognition, the computational equivalent of universal knowledge applied at scale.
But phronesis is something else entirely. It is the wisdom to know that a 78-year-old man with stage IV pancreatic cancer does not simply need to know his prognosis; he needs someone to sit with him, to understand what kind of death he fears, what his grandchildren mean to him, and whether treatment or comfort is the right path. No algorithm, however sophisticated, navigates that room.
Aristotle would recognise this gap immediately. He would applaud the machine's episteme while insisting that medicine, at its highest expression, demands phronesis, and that this remains irreducibly human.
Telos: What Is Medicine Actually For?
Aristotle was a teleologist. Everything has a telos, a purpose, an end toward which it strives. The acorn's telos is the oak. The eye's telos is to see. And the physician's telos, he would argue, is not merely to cure disease but to promote eudaimonia, human flourishing.
This raises a profound question about AI in healthcare. What is its telos?
If we deploy AI primarily to reduce diagnostic error, the telos is accuracy. If we deploy it to reduce cost and increase throughput, the telos is efficiency. If we deploy it to democratise access, bringing expert-level diagnostics to rural Nigeria or underserved communities in the American Midwest, the telos approaches something closer to justice.
Aristotle would insist we are explicit about which of these we are pursuing, because the telos shapes everything: how we design the system, what we measure, whom we serve, and whom we leave behind. An AI optimised purely for accuracy in wealthy hospital systems may perform catastrophically in low-resource settings where training data is sparse. An AI built to maximise efficiency may quietly deprioritise patients whose complexity makes them time-consuming.
He would not oppose AI in medicine. But he would demand we answer, honestly, what it is for.
Virtue, Character, and the Soul of the Clinician
In the Nicomachean Ethics, Aristotle argues that virtue is not merely a set of rules to follow but a disposition of character, a cultivated way of being. The virtuous physician is not simply one who follows protocols correctly but one who has internalised compassion, honesty, courage, and justice so deeply that they act from these qualities naturally, in even the most difficult moments.
Here is where AI presents a genuine civilisational risk that Aristotle would take seriously.
If AI handles diagnosis, documentation, triage, and follow-up, what happens to the cognitive scaffolding of clinical practice in the formation of clinical character? Medical students learn judgment partly through exposure to complexity, ambiguity, and failure. They develop empathy through thousands of hours at the bedside. They build courage by having to deliver difficult news without algorithmic assistance.
If the algorithm is always right, always faster, always more confident, do we risk producing clinicians who are technically competent but morally underdeveloped? Aristotle believed virtue is forged through practice. Remove the practice, and what remains?
This is not an argument against AI. It is an argument for designing AI integration into medical education and clinical training with enormous care, ensuring that the formative experiences that build physician character are preserved, even as the tools around them become more powerful.
The Four Causes of an AI Diagnostic System
Aristotle explained change and existence through four causes. It is an instructive exercise to apply them to something like an AI imaging system reading a chest X-ray:
Material cause: What is it made of? Data. Millions of labelled images, the crystallised observations of generations of radiologists encoded into weights and parameters.
Formal cause: What is its structure? A deep neural network: layers of mathematical transformations learning to detect edges, textures, patterns, and ultimately pathology.
Efficient cause: What brought it into being? The engineers, data scientists, and clinicians who built, trained, and validated it and the healthcare institutions that funded them.
Final cause: What is it for? Here, Aristotle would pause. The honest answer is that different stakeholders have different answers. For the radiologist, it is a second pair of eyes. For the hospital administrator, it is scalability. For the patient, it is, ultimately, survival and health.
The final cause, the telos, is the one that matters most. And in a world where AI development is shaped as much by commercial incentive as by clinical need, Aristotle would be asking hard questions about whose final cause is actually being served.
A Note on Hubris
Aristotle admired Sophia, wisdom, and distrusted its corruption into hubris, the overarching pride that mistakes capability for understanding. Ancient tragedy is full of it.
There is a version of AI in healthcare that carries this risk: the belief that because a system achieves 94% accuracy on a benchmark dataset, it understands disease. It does not. It recognises patterns. These are profoundly different things. The pattern recogniser trained on scans from one population may fail quietly on another. The language model that summarises a patient's notes may produce a plausible-sounding hallucination. The prediction algorithm may encode historical inequities into its future recommendations.
Aristotle would counsel epistemic humility. The machine knows what it has seen. Medicine requires wisdom about what it has not.
The Philosopher's Verdict
Standing near the birthplace of this philosopher, I imagine Aristotle encountering a modern hospital's AI ecosystem, the imaging algorithms, the clinical decision support tools, the robotic surgeons, the ambient documentation systems, with the same methodical curiosity he brought to octopuses and constitutions and the movement of celestial spheres.
He would not be afraid of it. He was never afraid of complexity.
He would ask: Does this serve human flourishing? Does it cultivate or diminish the virtues of those who practice medicine? Does it expand access to the good life, or concentrate it? Is it honest about what it does not know?
What is the ultimate purpose of AI in society? Efficiency? Profit? Automation? Human augmentation? Equity? Longevity? Creativity? Aristotle would insist that before asking what AI can do, society must ask what AI is for.
This question becomes even more urgent as AI moves from tool to decision-maker.
In surgery, AI-assisted robotics may augment dexterity and precision. In education, AI tutors may personalise learning. In public health, predictive systems may anticipate disease outbreaks before they occur. Yet the same technologies can widen inequality, amplify bias, displace workers, distort truth, and weaken trust.
Aristotle warned repeatedly against excess. Virtue, he believed, existed in balance, the “golden mean” between extremes. Perhaps this is the framework we need for AI today: neither blind optimism nor fear-driven resistance, but disciplined stewardship.
The future may not belong to those who simply build the most powerful AI systems. It may belong to those who most thoughtfully integrate human wisdom with machine capability.
He would remind us of what he wrote in the Nicomachean Ethics, that the purpose of knowledge is not knowledge itself, but the good life of human beings. The stethoscope, the scalpel, the MRI, and now the algorithm are all instruments in service of that end.
AI may surpass humans in computation, but civilisation will still depend on uniquely human judgement.
The question is not whether AI belongs in medicine. It already does, and will increasingly so. The question Aristotle would press upon us, the question worth carrying from this ancient Macedonian land into every hospital boardroom, every medical school, every regulatory chamber, is whether we are wise enough to use it well.
Aristotle taught us that wisdom is not the accumulation of knowledge, but the ethical application of it.
In the age of AI, that distinction may matter more than ever.
