Vincent Jacquelinet
CEO
Mourad Hassani
Head of R&D
4 minutes
For several months now, artificial intelligence has made its way into every debate. What was until recently a technical concept reserved for a few experts has spread in record time to the general public.
Just two months after its launch in late 2022, the first public version of an LLM-based AI (OpenAI's ChatGPT) had already reached 100 million users, a historical record for a new consumer technology. To reach this threshold, the Internet took 7 years, Facebook 5 years. (1)
By late 2025, just 3 years later, AIs based on LLM language models such as ChatGPT, Microsoft Copilot, Google Gemini, Anthropic's Claude, or Mistral are already used by more than one billion people every week.
The acceleration in the adoption of this new technology does not only affect the general public. It is also spreading to professional uses, including in sectors historically slower to adopt new technologies, such as healthcare.
In the United States: the American Medical Association pointed out that by early 2026, 81% of doctors reported using AI at work (compared to 38% in 2023) (2)
In Europe: the momentum is also underway. The European Commission, for example, indicated in 2024 that more than 60% of European hospitals were already experimenting with AI solutions. (3)
In France: adoption is already massive. According to the 2025 Data & AI in Healthcare Barometer, 90% of healthcare professionals report using artificial intelligence tools in their practice, with 32% doing so daily. More than half (53%) believe that AI has already changed the way they practice (4)
But this adoption, rapid as it may be, requires at least two clarifications: what are we really talking about when we talk about AI? And above all, what is actually happening with medical AI?
Healthcare: let's start with the ground, not the technology
To talk usefully about AI in healthcare, the most honest approach is to start from what patients and caregivers experience daily, before talking about models or algorithms.
On the patient side, access to a doctor has further deteriorated in recent years: the average time to get an appointment with a GP went from 4 days in 2019 to 12 days in 2026, which is 3 times longer (5)
On the caregiver side, the situation is also difficult: 59% of French people report having already given up at least one healthcare treatment over the last 5 years due to excessively long waiting times, which increases pressure on doctors and healthcare teams. (5)
It is in this context of strained access to healthcare and limited medical time that AI in healthcare deserves to be looked at. And the question is not simply to know whether it is technologically impressive. It is much more pragmatic: can AI help caregivers make better use of a time that has become scarce, without undermining quality, safety, or the patient-caregiver relationship?
AI technologies have reached a new milestone. Language models are now capable of summarizing, rephrasing, extracting information, structuring documents, or debating in natural language. Vision systems continue to progress in medical imaging, for example in radiology, ophthalmology, or pathology. Hybrid approaches, which combine statistical models and structured medical knowledge, are paving the way for more reliable and explainable uses.
The current acceleration of AI adoption in healthcare does not come from technology alone. It comes from the meeting point between a new technical maturity and a pressure on the healthcare system that is as strong as ever: longer appointment waiting times, medical deserts, complex pathways, administrative burden, difficulty in ensuring continuity of care.
However, we must avoid two pitfalls. The first would be to present AI as an automatic revolution, capable of solving the health system's pressures on its own. The second would be to reduce it to a passing fad. Between these two positions, there is a more useful space: looking at what is already working, what remains to be proven, and the necessary conditions for these tools to actually improve medical practice.
When we say "AI in healthcare", what exactly are we talking about in AI?
Not all AI, in healthcare or elsewhere, plays the same role. Some have a direct medical purpose: helping with screening, diagnosis, or therapeutic decisions. Others are support tools: writing summaries, structuring reports, helping with organisation, or automating administrative tasks.
This distinction is important because the regulatory requirements, expected levels of evidence, and risks are not the same. The HAS precisely distinguishes digital medical devices for professional use, for example for diagnostic aid, and technologies without a direct medical purpose used to reduce administrative burden. (Haute Autorité de Santé)
Behind the word "AI", several technical families coexist.
Symbolic AI relies on rules, decision trees, clinical scores, ontologies, or knowledge graphs. This approach offers major advantages: transparency and traceability. It is possible to understand why a conclusion is produced by the AI, follow its reasoning, and identify its limits. Humans can understand which rules were applied and why a conclusion was proposed.
Machine Learning and Deep Learning learn from data. They are particularly used to recognise patterns in images, signals, or structured records. They can perform very well on a specific task, provided they have representative, well-annotated data - meaning accompanied by a reliable response - that is adapted to the context of use.
Large language models, or LLMs, manipulate natural language. They can summarise a file, rephrase an explanation, extract information from a text, prepare a summary, or help write a report. Their strength is their flexibility; their limitation is that they can produce a convincing response without it being accurate.
RAG, for Retrieval-Augmented Generation, consists of connecting a language model to selected sources — guidelines, protocols, reference standards, patient records, or scientific literature — before producing a response. This does not make the response automatically true, but it reduces the risk of error and improves its factual grounding and verifiability.
Agents orchestrate several steps: querying a document database, applying a rule, extracting a data point, verifying information, and then producing a summary. In healthcare, their interest lies not in "reasoning alone", but in chaining together useful, controllable steps integrated into clinical practice.
AI Family | Header 2 | Header 3 |
|---|---|---|
🧠 Symbolic AI | 1950s-1980s | Uses rules, decision trees, ontologies, or knowledge graphs to explicitly represent knowledge. |
📊 Machine Learning | 1980s-2000s | Automatically learns from data to identify regularities and make predictions. |
🖼️ Deep Learning | 2010s | Uses deep neural networks capable of analysing images, sound, or large volumes of complex data. |
💬 LLMs (Generative AI) | Since 2022 (general public) | Understands and generates natural language using very large models trained on billions of texts. |
🔎 RAG & AI Agents | Since 2023 | Combine language models, documentary sources, and automation of complex tasks to perform more reliable and contextualised actions. |
In a few decades, artificial intelligence has evolved from systems capable of explicitly representing knowledge to models capable of learning from data, and then to systems capable of understanding, interacting, acting and explaining.
AI in healthcare does not refer to a single technology. It encompasses very different tools, with highly variable levels of risk, evidence and integration. And the most useful applications often arise from the combination of these building blocks — LLMs, data, structured knowledge and orchestration — rather than a single one. Aldebaran relies on hybrid AI in healthcare, blending these different tools.
In the next blog post: we will talk about the uses of AI in healthcare, its promises, but also its limitations.
(1) OpenAI. “The State of Enterprise AI 2025 Report.” 2025.
(2) American Medical Association (AMA). “Augmented Intelligence in Health Care.” 2026.
(3) European Commission. “AI in Healthcare.” Report, 2024.
(4) Cegedim Santé. “AI Health Barometer 2025: uses, risks and training.” 2025.
(5) French Hospital Federation (FHF), Ipsos and BVA. “Barometer on the uses of artificial intelligence in healthcare.”
