Lyon 3 · AI & Marketing Research Lab
Illustration of a young person pointing upward
A two-minute scroll

Medical AI
Literacy.

What it is.
What it cannot do.
What it means for your care.

Scroll

AI is already woven into health. Sometimes you can see it. Sometimes it works invisibly — and still shapes real outcomes.

Adapted from OECD/European Union (2026), p. 5.

Inside
Medical AI
Step 1 of 3

It reads your data

Your vitals, lab results, scans — and the notes your doctor typed about you.

An AI robot reading a medical chart
Whatever is missing from your file is invisible to it.
Inside
Medical AI
Step 2 of 3

It compares you

With millions of past patients. That is statistics, not understanding — it computes what is most likely.

A woman working with a neural-network diagram
It has never met you, or examined anyone.
Inside
Medical AI
Step 3 of 3

It suggests something

A risk score, a plain-language letter, an appointment sooner, a possible diagnosis.

A robot presenting a medical report
A human still decides. That human can be asked why.
The law calls all of this AI.

A machine-based system that … infers, from the input it receives, how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments.

Definition of AI in the EU AI Act (European Parliament, 2024), quoted in OECD/European Union (2026), p. 6.

Two families,
One creates.

Generative AI

creates new content based on existing materials in its training data, powering tools like chatbots, coding assistants and image generators.

What that means for you
discharge summary: pt afebrile, cont. abx 5/7, f/u PRN, amb. w/o assist

OECD/European Union (2026), p. 6, “Types of AI”.

Two families,
One predicts.

Predictive AI

is trained to detect patterns in existing data and to forecast what is likely to happen next.

What that means for you
30 past patients like you
0%chance of needing care sooner

It warns your team that you may need to be seen sooner than planned.

OECD/European Union (2026), p. 6, “Types of AI”.

A green AI head facing a human head
Keep this in mind
Keep this
in mind
It can
sound
caring.
It can sound certain.
It is not a clinician.

“AI systems use data to produce outputs that are statistically likely to meet an objective or reward. These systems do so without awareness, understanding or intent.”

0% of surveyed US teens had used AI “companions”. Patients bring symptoms — and fears — to the same chatbots.

OECD/European Union (2026), p. 7, citing Robb & Mann (2025).

A simple framework

Four moves.
One AI-literate patient.

Engage, create, manage, shape — four habits that keep you in charge.

Four overlapping shapes representing the four domains

Domains of the AILit Framework: OECD/European Union (2026), p. 8, Figure 1.

Move 1 of 4

Engage

“Become a critical and responsible participant in a world marked by AI”

Two people at a computer checking AI-generated results

Notice when AI is part of your care — and ask.

Competence: “Evaluate whether AI outputs should be accepted, revised or rejected.”

Move 2 of 4

Create

“Use AI as a creative partner while maintaining human agency”

Two people creating together with an AI tool on a tablet

Use it to prepare your questions, in your own words.

Competence: “Direct generative AI systems to elicit feedback, refine results and support reflection.”

Move 3 of 4

Manage

“Divide work intentionally between humans and AI”

A team dividing work with an AI assistant

Let it handle the paperwork. Keep the decisions human.

Competence: “Decide whether to use AI systems based on the nature of the task.”

Move 4 of 4

Shape

“Improve AI systems to reflect human values”

Two people examining an AI system on a large monitor

Say when it gets you wrong. Ask who it was trained on.

Competence: “Investigate how an AI system is intended to work, whom it is designed for and what its limitations are.”

Before you trust it

Three checks.
Every time.

Fairness, openness, your data. Three questions to ask when AI is part of your care.

A medical AI system being reviewed on screen
Check 1 of 3

Who is it fair to?

If it learned mostly from people unlike you, it can read you wrong — and no one says so.

A person and an AI system meeting on equal footing
Bias inherently exists in AI systems, which can also reflect societal biases embedded in training data or algorithm design.

OECD/European Union (2026), p. 20, Knowledge 2.5.

Check 2 of 3

Can you see how it works?

You are allowed to ask: what does it do, who was it built for, where does it fail?

An AI system showing how it reached its result
Ensuring transparency, explainability, and intelligibility.

WHO principle, in PAHO/IDB (2024), p. 38.

Check 3 of 3

Where does the data go?

Your medical record is the most personal data you own. Ask who sees it, and for how long.

A shield protecting personal health data
Privacy, confidentiality, and security of data use must be foundational to every AI development.

PAHO/IDB (2024), p. 37.

A globe with sparkles

Stay curious.
Stay in charge of your care.

“Embracing AI in public health is a collective effort to ensure no one is left behind.”

Dr. Jarbas Barbosa, Director, PAHO/WHO, in PAHO/IDB (2024).

Return to the survey

Illustrations and quoted excerpts are reproduced or adapted from OECD/European Union (2026), Empowering learners for the age of AI: An AI literacy framework for primary and secondary education, OECD Publishing, Paris, https://doi.org/10.1787/65cd27d4-en, licensed CC BY 4.0. This is an adaptation of an original work by the OECD and the European Union. The opinions expressed and arguments employed in this adaptation should not be reported as representing the official views of the OECD, its Member countries or the European Union. Illustrations by Abiyasa Adiguna. Health excerpts: PAHO/IDB (2024), Artificial intelligence in public health: Readiness assessment toolkit, Washington, D.C. Research stimulus for educational purposes; not medical advice.