Intelligence is not wisdom: preparing medicine for the age of coexistence (with machines and algorithms)
July 9, 2026
Josep Manel Picas: Innohealth Academy
A few weeks ago, the philosopher and AISOMA founder Murat Durmus published a short essay, The Philosophy of Coexistence, which poses a question that should unsettle all of us who work in health: how can we remain free, meaningful and morally responsible in a world where intelligence is no longer exclusively ours?
It is not an abstract question. It is the question behind every decision-support algorithm that enters a consultation, every detection system that accompanies an endoscopist and every language model that drafts a report. It is worth pausing on.
The distinction that changes everything
Durmus’s central thesis fits in a single line: intelligence is not wisdom.
Intelligence optimizes, answers and expands power. Wisdom, by contrast, asks what is worth optimizing, which questions deserve to be raised and how that power is kept under control. The first is instrumental; the second, reflective and oriented toward ends. It is an old distinction (Aristotle already separated theoretical knowledge from phronesis, practical prudence), but it has never been as urgent as now.
Durmus defuses a premise we often take for granted: that our dignity depends on being «the smartest thing in the room». If that were so, it would always have been fragile, because every more capable machine would be a small humiliation. Human worth, he argues, does not lie in calculation speed but in vulnerability, responsibility, mortality, care and the stubborn habit of seeking meaning. Precisely the competencies that define good clinical practice.
The real danger is not the machine
Here is the essay’s most interesting turn, and the one that speaks most directly to our sector. The main risk, Durmus says, is not an omnipotent AI, but a passive humanity: that we surrender judgment in silence, that we mistake fluency for truth and take comfort for destiny, while still calling «just a tool» something that has already begun to shape who we are.
A text generated by an AI can be flawless, coherent and persuasive without being correct, or understood by whoever produced it. Emily Bender and her colleagues named this the problem of «stochastic parrots»: form without meaning. It is the contemporary version of John Searle’s Chinese room: competence without comprehension. Computational sophistication does not amount to judgment, empathy or causal reasoning.
From philosophy to the consultation
This warning has a surprisingly literal clinical translation, and the evidence is starting to accumulate.
In August 2025, The Lancet Gastroenterology & Hepatology published the first study documenting a deskilling effect caused by clinical AI. In four endoscopy centres, the adenoma detection rate of professionals working without AI assistance fell from 28% to 22% after just three months of routine use of the tool. Put another way: by delegating vigilance to the algorithm, specialists lost part of their own sharpness. (With caution: the volume of colonoscopies nearly doubled over the same period, so other factors may be at play; but the signal is clear enough to take seriously.)To this we add automation bias: the well-documented tendency to favour the machine’s suggestion even when it is wrong. Recent studies in computational pathology show that under time pressure professionals go so far as to reverse initially correct assessments in order to follow a mistaken AI recommendation. The passivity Durmus fears is not a philosophical hypothesis: it is a measurable phenomenon at the bedside.
It is no coincidence that the World Health Organization, in its guidance Ethics and Governance of Artificial Intelligence for Health, sets as its first principle that humans must retain control of health systems and medical decisions. The human-in-the-loop model, systems designed so the clinician reviews, interprets and, where necessary, overrides the recommendation, is not a bureaucratic formality: it is the concrete safeguard against deskilling, the never-skilling of new generations and technostress.
What it means, in practice, for innovation in new models of health
If intelligence can be bought and scaled, wisdom has to be cultivated. And that is, at bottom, a matter of training. Some principles follow:
- Train judgment, not just tool use. Teaching how to use an AI co-pilot without teaching how to doubt it is training professionals vulnerable to automation bias. AI literacy must include calibrated skepticism.
- Preserve “unassisted” practice. If evidence shows that skill atrophies with continued delegation, deliberate spaces for unassisted practice must be designed, like a muscle that must be trained.
- Make responsibility explicit. The “human in the loop” only works if the clinician knows that the decision—and its justification—remains theirs. Convenience cannot become abdication.
- Cultivate what the machine lacks. Shared vulnerability, presence, listening, causal reasoning in the face of the atypical case. These are not “soft skills”: they are the irreplaceable core of care.
The question the future will ask us
Durmus closes his essay with a line we should pin up in every lecture hall and boardroom: the future will not ask us whether we were the most intelligent beings in the room, but whether we were wise enough to live alongside an intelligence greater than our own.
The coexistence he proposes is neither surrender nor panic. It is humility without renunciation and responsibility without fear. For an ecosystem like healthcare, where every decision has a body, a name and a biography in front of it, this attitude is not a philosophical luxury.
It is, increasingly, a professional competence.
Some references and further reading
- Durmus, M. (2026). The Philosophy of Coexistence: An Attempt to Expand Philosophy for the Age of Artificial General Intelligence. (Butlletí Thoughts on AI, LinkedIn, 5 de juliol de 2026.)
- World Health Organization (2021). Ethics and Governance of Artificial Intelligence for Health: WHO Guidance.
- Bender, E. M., Gebru, T. et al. (2021). «On the Dangers of Stochastic Parrots». FAccT ’21.
- Searle, J. R. (1980). «Minds, Brains, and Programs». Behavioral and Brain Sciences.
- Vallor, S. (2024). The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking. Oxford University Press.
- Parasuraman, R. i Manzey, D. H. (2010). «Complacency and Bias in Human Use of Automation». Human Factors.
- Simposi InnoHealth Academy 2025: Conclusions · innohealth.academy · www.innohealth.academy
