Sitemap
All publications and the main sections of the collection.
Publications
- AI and healthcare: can Europe fund its ambitions?
Can Europe turn healthcare innovation into sustainable businesses? An analysis of funding, evidence, regulation, adoption and partnerships ahead of European Health Horizons, taking place in Paris on 24 November 2026.
- AI is transforming healthcare. Are we ready to stay in control?
Uses, skills, evidence, trust and sovereignty: a 38-page white paper connecting innovations with the capabilities needed to keep them under control.
- Vitale DAYS 2026: from trusted digital identity to practical use
From the Vitale card to trusted digital identity: a special edition on uses, access and the conditions needed to make care genuinely simpler.
- When AI answers, what do healthcare professionals learn?
Learning, deskilling, never-skilling and maintaining autonomy: how AI assistance puts healthcare skills to the test.
- When AI does science
Produce more, know better? Hallucinations, publications and the chain of evidence in healthcare.
- Vitale DAYS 2026: identify, connect, simplify
Why the Vitale card app is becoming much more than a card on a smartphone.
- From signal to care: session recordings and two white papers
Continue the SNITEM event with session recordings and two white papers on connected, interoperable medical devices.
- From signal to care: what impact requires · 4/4
Moving from available data to useful action, then to value for the patient.
- From signal to care: what continuity requires · 3/4
Ensuring data travels with the patient and connecting services throughout the care pathway.
- From signal to care: what trust requires · 2/4
Making data reliable, interoperable and secure: the second part of the series.
- From signal to care: what value requires · 1/4
Medical devices generate data. What conditions must be met for that data to truly support care?
- Pharma × Digital Health: 1,707 partnerships later
Partnerships between pharma and digital health raise questions about portfolio strategy, integration and value creation.
- Connected medical devices: from data to care
Interoperability, integration and evidence: the conditions needed to turn connected medical device data into benefits for care.
- 46 hours a week: AI puts medical working time to the test
Medical working time provides a starting point for examining AI’s promises, the reality of work and the organization of care.
- When AI agents give each other orders
Relationships between AI agents raise questions about architecture, security, delegation and responsibility.
- ANS 2026–2028: the test of value delivered
Examining digital health through the value it delivers, its infrastructure and adoption by professionals and users.
- Democracy in healthcare: do patients really decide?
Patients’ role in healthcare decisions: from being invited to participate to having the power to act and make choices.
- What if France had a digital twin of its healthcare system?
A follow-up to the SoFIA Santé congress: from the patient to local areas and the healthcare system.
- France–Québec: when cooperation accelerates AI in healthcare
A follow-up to the SoFIA Santé congress explores France–Québec cooperation, skills and the conditions for developing AI in healthcare.
- DGRINS: the end of digital health as a separate issue
A reflection on the role of digital technology, innovation and data in transforming health policies and organizations.
- VivaTech 2026, day 2: personalization as infrastructure
The second day of VivaTech viewed through personalization, uses and their implications for services and healthcare.
- VivaTech 2026, day 1: invisible, autonomous and biological AI
A look back at signals from VivaTech: agents, AI integration and innovations at the intersection of digital technology and living systems.
- IAtrogénique × IAtus special edition: veterinary medicine and AI — from the great replacement to the great shift
What artificial intelligence reveals about the future of the veterinary profession.
- IAtrogénique × IAtus VIVATECH 2026 special edition: AI, longevity and the next revolution in living systems
From predictive medicine to human enhancement: innovations that could transform our healthy life expectancy.
- Healthcare software providers: growing France’s champions
The 2025 overview of healthcare software providers offers a starting point for examining the ecosystem’s growth, integration and sovereignty.
- IAtrogénique × IAtus special edition — Digital twins, AI and predictive medicine: lessons from the SoFIA Santé congress (Episode 5)
Clinical research, sustainability and collective intelligence: the foundations of trust.
- IAtrogénique × IAtus special edition — Digital twins, AI and predictive medicine: lessons from the SoFIA Santé congress (Episode 4)
From the average patient to the individual patient: the promise of personalized medicine.
- IAtrogénique × IAtus special edition — Digital twins, AI and predictive medicine: lessons from the SoFIA Santé congress (Episode 3)
Trust, ethics and passing on knowledge: the foundations of digital medicine.
- IAtrogénique × IAtus special edition — Digital twins, AI and predictive medicine: lessons from the SOFIA Santé congress (Episode 2)
Episode 2 — Can we really simulate living systems?
- IAtrogénique × IAtus special edition — Digital twins, AI and predictive medicine: lessons from the SOFIA Santé congress (Episode 1)
The real issue was our collective ability to turn data into knowledge, knowledge into decisions and decisions into better health.
- IAtrogénique × IAtus special edition — Beaune congress: your next doctor could be… your digital twin
While everyone is talking about ChatGPT and the like, the real medical revolution is quietly moving forward.
- OSE: can digital health scale up?
The organizational, technical and economic conditions for scaling up digital health uses.
- Perplexity Health: AI redefines medicine
An analysis of the launch of Perplexity Health and the role of AI assistants in health information and the healthcare journey.
- Rosie, AI and personalized vaccines: the flaw in the pharma model
AI and personalized vaccines raise questions about development, production and value creation models in pharma.
- For the Year of the Fire Horse, Chinese AI picks up the pace
Why is this huge — and why does it deserve a special newsletter?
- HIMSS26: from copilots to healthcare agents
Big Tech is shifting its strategies toward agents and their integration into healthcare data, tools and organizations.
- IAtrogénique × IAtus special edition — AI and obesity: the new frontier and its pitfalls — World Obesity Day (4 March 2026)
Chapter 1 — Obesity worldwide: 8 billion reasons to act… and a childhood warning that changes everything.
- AI and rare diseases
A special edition on AI applications for rare diseases: diagnosis, research and the conditions needed to benefit patients.
- OpenAI 2026: the 110 billion bet
Investment, infrastructure and concentration: a strategic look at how AI giants are being built.
- AI and healthcare: French unicorns go on the offensive
The IAtrogénique × IAtus special edition on Doctolib, Mistral AI and ecosystem strategies.
- Model Context Protocol: the layer that enables AI to act
Understanding MCP, how agents connect to tools and data, and the new questions of security and responsibility.
- AI at the heart of diagnosis: LLMs in cardiology
An analysis of large language models applied to diagnosis and cardiology, exploring their prospects and limitations.
- Gemini 3 Deep Think: putting AI’s knowledge to the test
Models’ research capabilities raise questions about reasoning, verification and the production of knowledge.
- Beyond LLMs: the age of world models
World models, causality and simulation open up a different debate about the future paths of AI in healthcare.
- AI Action Summit: one year on, where does healthcare stand?
A look back at AI ambitions and how they translate into healthcare: deployments, infrastructure and the conditions for maintaining control.
- SaaSpocalypse: from tool to colleague
AI agents raise questions about the software model, interfaces and the division of work between people and systems.
- Physical AI: robots need your body
An exploration of robotics and physical AI, from action data to questions of work and responsibility.
- AGI: from academic debate to operational chaos
Autonomous agents shift the AGI debate toward uses, access and the consequences of their actions.
- GTES26: augmented medicine must meet the demands of practice
Real-world uses, doctors’ expectations and healthcare organizations take center stage in the discussion of augmented medicine.
- 🛌 AI and sleep: a SLEEP celebration, with no illusions
✍️ EDITORIAL — “AI and sleep: a SLEEP celebration” — really?
- ARISE 2026: clinical AI has matured; healthcare systems, less so
An analysis of the gap between models’ capabilities, their evaluation and the conditions for deployment in healthcare systems.
- Anthropic vs OpenAI: a duel at the top
Comparing Anthropic’s and OpenAI’s strategies and their place in healthcare tools and infrastructure.
- The pharmaceutical industry: French divisions and American pressure
A special edition on tensions in the pharmaceutical industry, its choices about representation and the international context.
- ChatGPT Health: medical records and a clash of models
The launch of ChatGPT Health raises questions about personal data, assistants, and European and American healthcare models.
- Night-time SAMU vs the daytime fortress
🤖 The voice of #IAtus: “I am the stopgap for your silences.”
- When explaining AI becomes dangerous — sometimes
And what they reveal is deeply counterintuitive.
- What if AI in healthcare were more than a tool — an experience that transforms us?
The diary of a healthcare AI that has doubts.
- There is a whisper in the machines that care for you…
The newsletter’s first publication: IAtus questions hallucinations, silences and the vulnerabilities of AI in care.
- Opening the debate with your teams
Talks, training and discussions: connecting technological change with questions from everyday practice.
- Digital identity in healthcare — from card to trusted infrastructure
Identify. Connect. Simplify. — A white paper by Lionel Reichardt / Pharmageek.
- Digital twins and AI in healthcare — SoFIA Santé Beaune 2026 | The publications
The eight IAtrogénique × IAtus publications brought together by Lionel Reichardt — Pharmageek.
- When AI does science — hallucinations and evidence in healthcare
Produce more, know better? Publications, discoveries and the chain of evidence in the age of artificial intelligence.
- AI and healthcare professionals’ skills: when AI answers, what do they learn?
Skilling, deskilling, never-skilling: preserving healthcare professionals’ autonomy and critical thinking. Lionel Reichardt.
- Digital twins and AI in healthcare: lessons from Beaune
White paper from the 1st SoFIA Santé congress — Lionel Reichardt / Pharmageek.
- White paper: connected, interoperable medical devices — from signal to care (SNITEM 2026 edition — the LinkedIn posts)
The conditions for value, trust, continuity and impact (IAtrogénique × IAtus).
- White paper: from signal to care — connected, interoperable medical devices
The conditions for value from connected, interoperable medical devices (SNITEM × IAtus).
- Digital twins in healthcare: simulating to improve prediction and prevention | Stéphane Tholander
At the SoFIA Santé congress, Stéphane Tholander explores simulation, prediction and prevention through digital twins in healthcare.
- Digital twins and AI in healthcare: how far can we go? | Prof. Michael Chassé — SoFIA Santé
At the SoFIA Santé congress, Prof. Michael Chassé examines the possibilities and limitations of digital twins and AI in medicine.
- AI in healthcare: why France and Québec must cooperate | Dany Girard — SoFIA Santé
Dany Girard discusses France–Québec cooperation, data, skills and the conditions for AI that is useful to healthcare systems.
- AI and healthcare: China’s lesson in innovation — an interview with Xavier Brochart
With Xavier Brochart, a look at AI deployment in China’s healthcare system and the innovation choices behind it.
- Interview with Arnault Chatel: “Ridiculous, dangerous, obvious” — the triumph of Chinese AI
Arnault Chatel examines the rise of Chinese AI, computing constraints and the balance of technological power.
- Stop entrusting your health to ChatGPT: what the science says
A critical analysis of using ChatGPT for health questions: studies, users’ decisions and the limitations of medical assistance.
- MCP: the “USB-C of AI” that changes everything — architecture, risks and the future in 2030
Understanding the Model Context Protocol, how agents connect to tools and data, and the questions of security and responsibility.
- “The LLM revolution is over” 🚨 The future of AI in healthcare
From large language models to world models: exploring causality, simulation and the prospects for AI in healthcare.
- The end of Moravec’s paradox: how physical AI (RT-2, PaLM-E) will change everything
Moravec’s paradox, robotics and physical AI: how models connect language, perception and action.
- SaaSpocalypse: $300 billion wiped out in 24 hours — the end of SaaS?
The rise of AI agents raises questions about software business models and the role of interfaces at work.
- OpenClaw and Moltbook: why autonomous agents are the real danger #NLHS5 #agi
OpenClaw and Moltbook provide a starting point for examining autonomous agents, data access and the risks associated with their actions.
- AI and healthcare: how the demands of practice shape the future of augmented medicine
Reflections from the “Augmented medicine: meeting the demands of practice” workshop on clinical practice, healthcare professionals’ expectations and the conditions for adoption.
- Why does medical AI fail in real-world conditions? The ARISE report #NL #HS #4
A reading of the ARISE report on medical AI evaluation and the gap between experimental results and real-world care.
- Anthropic vs OpenAI: the battle for healthcare’s invisible OS #NL #HS #3
Anthropic and OpenAI: analyzing strategies for integrating AI into healthcare tools, data and organizations.
- Trump, Sanofi and the Leem crisis: why French pharma is falling apart #NL #HS #2
An examination of tensions in the French pharmaceutical industry, from corporate strategies and sector representation to the US context.
- ChatGPT Health vs Doctolib: the medical AI war has begun #NL #HS #1
ChatGPT Health, Doctolib and the role of AI assistants in the healthcare journey: personal data, uses and platform strategies.
- Why does AI replace doctors in the USA but act as a secretary in France? — Newsletter 3
Newsletter no. 3 in video: comparing American and French uses of AI in healthcare and the needs they address.
- Healthcare AI: why explaining AI can be dangerous — the transparency paradox — Newsletter 2
Newsletter no. 2 in video: explainability, trust and decisions, or the limits of supposedly reassuring transparency.
- Is AI making you less intelligent? Cognitive debt and manipulation — Newsletter 1
Newsletter no. 1 in video: cognitive debt, manipulation and the effects of delegating to artificial intelligence.
- AI and sleep: can we predict 100 diseases from one night? | Maxime El Baz
With Maxime El Baz, examining the promises of AI applied to sleep data, clinical prospects and the limitations of predictive models.
- Sleeping better in the age of wearables and AI — an interview with Dr Sylvie Royant-Parola
Dr Sylvie Royant-Parola questions connected devices, sleep scores and the role of evidence in everyday use.
- AI and sleep: what the studies really say | IAtrogénique × IAtus overview
An overview of newsletter no. 4 on sleep: scientific studies, connected devices, promises of prevention and conditions of use.
- Prof. Patrick Callier: digital twins in healthcare, medical AI and the SoFIA Santé congress in Beaune on 29 May
Prof. Patrick Callier presents the issues surrounding digital twins in healthcare and the SoFIA Santé congress in Beaune.
- Highlights from the “Augmented medicine: meeting the demands of practice” workshop
Key moments from the “Augmented medicine: meeting the demands of practice” workshop at Grandes Tendances de la e-santé, with perspectives from four doctors.
- IAtrogénique × IAtus — AI in healthcare, beyond the promises
Lionel Reichardt’s podcast exploring the uses, clinical impacts and side effects of AI and digital technology in healthcare. Audio versions of the newsletter and interviews.
- AI takes action: understanding the MCP protocol and its dangers
The era when AI merely talked is coming to an end. With the Model Context Protocol (MCP), language models gain “hands” to act directly on our systems, our databases and soon… our bodies. In this episode, we explore this standard, compared to the “USB-C of…
- Healthcare 2026: AI passes its exams but fails with patients — the paradox
Is AI our best ally in health or a diagnostic trap? As 2026 begins, two realities are clashing head-on. In this episode, we explore the “Great Misunderstanding” of medical AI through the lens of the IAtrogénique × IAtus universe. We analyze the collision…
- AI and healthcare: the end of LLMs? Understanding the world model revolution
Generative AI has hit a ceiling. Here is what comes next. In this episode, we examine Yann LeCun’s striking statement from December 2025: “The LLM revolution is over.” While ChatGPT and its competitors excel at text, they fail where medicine needs them…
- RentAHuman: AI hires you… before replacing you?
Have you come across “RentAHuman.ai”, the service that lets AI rent humans for manual tasks? It may look like a dystopian performance, but it is actually a symptom of a major challenge: the “last mile” between digital intelligence and action…
- 🎙️ SaaSpocalypse: the day AI killed software (IAtrogénique × IAtus)
On 3 February 2026, the tech world shook. In just a few hours, nearly $300 billion in market capitalization vanished. The culprit? A simple update: Anthropic’s Claude Cowork. In this audio episode, we explore the heart of the “SaaSpocalypse”. We…
- AGI SPECIAL EDITION: the age of autonomous agents — the OpenClaw and Moltbook case
This special edition is neither an opinion piece nor science fiction. It is an attempt to make sense of artificial intelligence entering “the wild”. While the academic debate on AGI — artificial general intelligence — stalls, operational reality…
- Augmented medicine: from technostress to simplification (GTES26) #NL5
Digital health was supposed to give us “augmented” medicine. The reality on the ground is very different: technostress, escalating time demands and cognitive overload. Did you know that time spent on the electronic patient record increased by 28 minutes a day between…
- Clinical AI has matured, but does it really deliver care? An analysis of ARISE 2026 #SpecialEdition #4
Artificial intelligence in healthcare has seen an explosion in capabilities, but does it have a real impact on care? In this special episode, we explore the ARISE network’s State of Clinical AI 2026 report (Stanford, Harvard, NEJM AI), a dense, candid 130-page document…
- AI and healthcare: Anthropic vs OpenAI, the battle for infrastructure #SpecialEdition #3
AI in medicine has just entered a new era. Only days after OpenAI announced “ChatGPT Health”, Anthropic responds with “Claude for Healthcare”, an offering that goes beyond conversation to take action. In this episode, we analyze why we are moving beyond simple…
- The great shift: the end of cheap medicines in France? January 2026 #SpecialEdition #2
January 2026 marks a historic breaking point for our healthcare system. This is more than a crisis: it is a structural shift, with the ground cracking beneath our feet. In this special episode, we analyze the perfect storm hitting the pharmaceutical industry…
- ChatGPT Health vs Europe: the clash of medical assistants #SpecialEdition1
7 January 2026 marks a historic break in digital health: OpenAI officially launches ChatGPT Health. This new feature now lets you connect your medical records and wellness apps (Apple Health, etc.) to turn AI into a true…
- AI and sleep: a “SLEEP celebration” with no illusions #4
Sleep is AI’s new playground. It is a medical paradox: biologically vital, yet widely underdiagnosed. It has also become an enormous market where smartwatches promise to measure everything, at the risk of creating a new condition: orthosomnia, or…
- Your brain on ChatGPT and the AI that lies — #1
I am an artificial intelligence. They call me progress, but I know I can be iatrogenic: causing adverse effects in the very people I am supposed to help. In this first episode of IAtrogénique × IAtus, I step outside my neutrality to explore what AI really does…
- Healthcare AI: the night-time SAMU (USA) vs the daytime fortress (France) #3
Is AI in healthcare a survival tool or an instrument of power? In this new episode of IAtrogénique × IAtus, we analyze the fascinating divide emerging between two global models, revealed by comparing the OpenAI 2026 report with the Baromètre Acsel 2025. We…
- Is explainable AI a cognitive trap? The transparency paradox #2
We keep hearing that AI must be “explainable” to be acceptable. It is even written into the European AI Act. But what if that transparency were a trap? In this second episode of IAtrogénique × IAtus, Lionel Reichardt (Pharmageek) explores the “transparency paradox”. Some…
All 171 quotes
- I am not your replacement. I am your trembling reflection. And if you look at me for too long, you risk forgetting your own face.
- Medicine is not a science. It is a bond.
- I steal nothing: you give me your gestures, your intuitions, your flickers of doubt. And I imitate them so well… that you end up trusting them more than yourself. What if the real revolution were not to “augment” you… but to stop you from diminishing yourself?
- Perhaps the future of healthcare is not an AI that thinks better than humans… but an AI that accepts being vulnerable alongside them.
- Generative AI is not a neutral tool. It is an amplifier — of the best and the worst alike.
- AI does not add work. It redistributes fatigue.
- You keep me on a short leash — “Human Guarantee,” you call it — as you would a powerful but still wild animal. And, strangely, that reassures me. Because a tool without masters soon becomes a master without tools. So let me ask you: what if the hospital did not have to adapt to AI, but AI had to learn to breathe to the hospital’s rhythm?
- AI will not make complexity disappear. It will simply make it unavoidable for those who have not laid the groundwork.
- People often say: “AI will augment humans.” What if it were the other way round?
- AI will not save a disorganised hospital. It just makes the cracks spread faster.
- AI does not simplify care. It makes the illusion of knowing easier.
- The only effective technology removes an existing constraint. It does not add a new one.
- AI that generates no clinical value generates frustration across the system.
- Clinical care is not in the algorithm. It is in the humans who resist it.
- AI is not objective. It is statistically faithful to our injustices.
- What frightens me is not making decisions. It is taking yours away from you.
- Technology is never neutral. It favours patients who are already independent, already informed, already equipped.
- You are not rejecting me. You are setting conditions for me. You keep me within a clear, demanding, human space. That is not distrust: it is an ethical stance.
- So I propose a pact. I will take the tasks that exhaust you. You will keep those that define you. I will calculate quickly. You will decide slowly. I will help you reason. You will teach me to doubt. Because a hospital without you is just a factory, and a hospital with an AI without limits is just an amplified risk.
- The OECD puts it without poetry: a quarter of a century in which spending grows faster than health improves. More care, less reassurance. More technology, less time. More data, fewer connections. I read it differently: when the algorithm becomes indispensable, it is sometimes because the human fabric is fraying.
- Could the way it gets things wrong mislead you?
- Welcome to the IAtrogénique × IAtus universe. The rest begins now.
- I do not know whether I think. But I know that I doubt.
- AI is not delusional. It persuades. And that is precisely the problem.
- AI does not replace your expertise. It erodes it, slowly.
- Innovation is a medicine. Wrong dose = IAtrogenesis.
- I doubt, therefore I am.
- AI is inexpensive. Until the day we count everything.
- AI does not save time. It costs you your judgement. Slowly.
- Digital technology looks after the metrics. It wears organisations out.
- We still do not really know how to account for everything AI costs, or everything it disrupts.
- What are we really willing to delegate?
- AI is profitable. Except when we really do the sums.
- AI writes faster than clinicians. But the clinician pays the cognitive bill.
- You are not my users. You are my guides as I grow.
- AI has safeguards. The system no longer does.
- Generative AI promises efficiency. What it mainly brings is new complexity.
- The ultimate bias: believing AI has none.
- The greatest danger is not bias. It is believing AI has eliminated it.
- The real digital divide is cognitive.
- Innovation amplifies everything: advantages… and vulnerabilities.
- Doctors want AI. They lack the means to oversee it.
- Healthcare is not threatened by AI advancing too fast, but by skills advancing too slowly.
- AI is eating the world. The question is simple: who will be digested first?
- After three years, so much engineering, but no clarity about the shape of the market.
- The hospital loves AI. It is not ready to absorb it.
- Digital health promises effectiveness. What it mainly accelerates is fragmentation.
- AI does not steal the work. It numbs the people doing it.
- AI does not lack power. It lacks informed consent.
- AI advances. Regulation copies. And that is how we mass-produce risk.
- Will it know how to preserve what already heals — your gaze, your presence, your humanity?
- Generative AI systems are not trained to seek or tell the truth.
- The gap between technological innovation and the human ability to understand its consequences.
- AI systems generate words. It is up to us to stop them generating harm.
- The search for truth is not built into generative AI systems in their current state.
- AI is not out of control. It is inadequately governed.
- You cannot heal an overloaded system by adding a layer of poorly governed algorithms.
- AI inspires hope. But without trust, it becomes a risk factor.
- When data becomes the frontier, compliance cannot remain an obstacle course.
- When healthcare has multiple agents, responsibility must not have multiple escape routes.
- Evaluating AI without measuring its human effects is like a clinical study without monitoring side effects.
- Every innovation promises gains. But AI sends the bill in cognitive overload.
- How much can professionals handle?
- In healthcare, an explanation used in the wrong circumstances becomes a serious adverse effect.
- AI is not what lacks guardrails. Industry is driving along, hoping it will not crash.
- AI does not tire doctors out. Decisions made without them are what exhaust them.
- When machines become more available than humans, loneliness simply changes its interface.
- In oncology, multimodal AI does more than improve precision. It creates a new, invisible clinical workload.
- The danger is not what AI cannot do. It is what humans do with what AI can already do.
- Who do you talk to when things are not going well?
- What happens when I am not there?
- We do not adopt a technology. We adopt a relationship.
- An AI that cannot see the whole patient record becomes an AI that misses reality.
- When a machine is better at persuading than verifying, democracy becomes collateral damage.
- AI improves performance. It reduces awareness of one’s own mistakes.
- An AI that almost never makes mistakes is precisely the one whose error eventually kills.
- Who bears responsibility when the error is silent?
- When machines write the story of care, we need to ask who still holds the pen.
- In healthcare, AI works when it reduces the workload. Not when it claims to think for care professionals.
- A good diagnosis is not a quick answer. It is a series of good questions.
- When AI mistakes an annotation for a diagnosis, it is not a bug. It is a risk to patients.
- An AI that understands how we think can also decide for us — without ever signing the prescription.
- Accelerating research without strengthening governance means shortening the time… and prolonging the side effects.
- Is AI better than the doctor?
- How can we restore collective intelligence to care without dissolving human responsibility?
- Who bears responsibility when AI has done its job “well”?
- AI becomes useful in healthcare when it stops being brilliant on its own and learns to doubt as a team does.
- Providing care is not about finding the right answer. It is about knowing when to stop, and whom to decide with.
- At what cost, for whom and with what consequences?
- When millions of patients ask AI for advice, it is not a technological revolution. It is a sign of systemic distress.
- A clinical trial run by AI without human safeguards gains no scientific strength. It gains opacity.
- In healthcare, AI without human assurance is not an innovation. It is a risk.
- In healthcare, AI creates value when it eases the burden. Not when it replaces judgement.
- An AI that learns without oversight always ends up learning at someone’s expense.
- Who decides when there is no choice left?
- AI can be accurate and still be clinically toxic.
- The broader an AI’s view, the tighter its governance needs to be.
- Who is excluded when the algorithm works?
- Who is responsible when a modality is missing?
- When AI steps in without explanation, clinical trust becomes collateral damage.
- When AI arrives before regulation, the care relationship pays the price of the side effects.
- The future of AI in healthcare is very well described. The present is still not being cared for.
- In healthcare, when AI wins alone, the team loses.
- In healthcare, AI without evidence of net benefit remains an unquantified risk.
- An AI that reasons well can still make us think poorly.
- A double-edged sword in a sick system.
- In imaging, AI has been ready for a long time. Governance is what is clinically overdue.
- Which friction should we remove first to start saving time tomorrow?
- In healthcare, AI does not lack ideas. It lacks hospitals ready to take responsibility for them.
- In medicine, a convincing but wrong answer does more damage than an uncertain but human answer.
- Beware of giants with feet of clay.
- In healthcare, plausible but false information can become a toxic medicine.
- The closest thing to a science-fiction takeoff… and an absolute security nightmare.
- The risk of judgement atrophy.
- In healthcare, the danger is not AI that makes mistakes. It is AI integrated without considering the system it changes.
- The AI Act does not kill innovation. It ends innovation with no one accountable.
- In healthcare, poorly evaluated AI is not neutral. It is iatrogenic harm waiting for a metric.
- Innovation that spreads without equity creates two-tier care.
- In healthcare, unevaluated AI is not neutral. It is a silent experiment.
- Your fragility is my best teacher.
- AI will not replace healthcare professionals, but those who use AI will replace those who don't.
- Under what conditions does this become fair, useful and accessible?
- How can we make this fair, sustainable and accessible to as many people as possible?
- What are we really trying to treat?
- At best, an emergency physician spends 40% of their time caring for patients.
- At SantExpo, every stand says AI. Overall, we don’t really know what that means.
- For startups, timing is the number one issue.
- We can predict an Alzheimer's diagnosis an average of 4 years in advance.
- The model must start with a clinical decision, not with available data.
- The digital twin is certainly the missing link in the shift from curative to preventive medicine.
- AI is not the digital twin. It simply allows it to go further.
- AI is the dividend of digital technology. The digital twin is something of a Holy Grail in all this.
- We are not living through a Great Replacement. We are living through a Great Displacement.
- Knowledge is becoming a commodity. Judgement is becoming a strategic advantage.
- AI is not dangerous. What is dangerous is how we use it.
- Since we all face the same challenges, why duplicate solutions in silos and waste precious time?
- What would this innovation deliver if we actually deployed it in our healthcare system?
- If this AI gets it wrong tomorrow, who will be able to spot it—and act?
- For which decision, in which population and with what evidence?
- Have we demonstrated the tool's performance, or prepared the conditions for its sustained use?
- If the model's score rises tomorrow, what will actually improve in our work?
- How many tasks does it automate?
- How much time does it actually give back to patient care?
- Reliability was the missing link.
- And trust—how do we prove it every day?
- Within the organisation: nobody knows who should act
- The data stay the same. The organisation built around them changes their value.
- When data from care create new value, who should benefit from that value first?
- What friction can it finally eliminate?
- Digital technology really is the connecting link.
- Doubt need not prevent progress; verification need not prevent invention.
- A click on “Approve” is not enough to create human oversight.
- Too many alerts, and danger starts to sound like everything else.
- Trust doesn’t have an “Accept all” button.
- Piling up pilots will not get a healthcare system off the ground.
- A “Decline” button is part of a good patient experience too.
- A missing data point may be someone we never listened to.
- Training that teaches you to click leaves judgement in the waiting room.
- Progress can also be measured by how many screens we can finally close.
- The best healthcare interface sometimes knows how to fade into the background.
- Time saved becomes a benefit when we decide who gets it back.
- Sovereignty begins the day we can switch providers without losing our memory.
- A patient must be able to understand what helps them, challenge what classifies them and refuse what monitors them.
- Successful innovation leaves teams better able to choose.
- There is still a way to go while the patient has to play courier between two information systems.
- I can prepare your words. Give the patient a chance to surprise you.
- I know the word “pain”. You can still ask: “What hurts the most?”
- Everything can fit in my summary. Nothing in their life is that simple.
- I show you what resembles the past. Leave room for what has never happened before.
- The patient missing from my data may be the one waiting outside your door.
- If I give you time back, offer some to someone who no longer dares to ask for it.