AI Literacy Is Not a Curriculum. It Is a Capability That Has to Keep Moving.

AI Literacy Is Not a Curriculum. It Is a Capability That Has to Keep Moving.

 

We are starting to make the same mistake with AI Literacy that we made with digital skills years ago.

We are turning it into a curriculum.

A few modules on how AI works.
A lesson on prompting.
A section on bias and hallucinations.
A checklist explaining what employees or students should and should not do.

Then we call people “ AI literate.” I do not think that is enough.

The problem is simple : AI is evolving faster than any static curriculum can keep up with it.

Models change. Their capabilities change. Interfaces evolve. New risks appear. New use cases emerge. What was considered a limitation six months ago may disappear, while entirely new questions around privacy, autonomy, trust and human responsibility emerge.

So AI Literacy cannot be treated as a body of knowledge that we teach once and consider complete.

It has to become a capability that continuously evolves.

 

Knowing about AI is not the same as being AI literate

Someone can explain what a large language model is and still trust an incorrect answer.

They can know the definition of a hallucination and fail to recognize one in practice.

They can complete a lesson on prompting without understanding how changing context, instructions, sources or even the model itself can radically alter an answer.

They can know that bias exists without knowing how to identify it and they can memorize an organization’s list of approved and prohibited AI uses while still lacking the judgement required when they face a situation that is not on the list.

This is why real AI Literacy develops through practice, experimentation and judgement. People need to :

• experiment with AI ;
• compare models ;
• question outputs ;
• verify claims against trusted sources ;
• recognize hallucinations and bias ;
• change context and observe the consequences ;
• understand how grounding affects a response ;
• identify privacy and security risks ;
• decide when AI adds value ;
• recognize when human judgement should remain primary.

This is very different from simply completing an AI Literacy curriculum.

The objective is not only to know what AI does. The objective is to become capable of working with it intelligently, critically and responsibly.

 

Perhaps the most important AI skill is knowing when not to use AI

AI Literacy is often framed around better use.

How do I write a better prompt ?
How do I automate this task ?
How can I generate content faster ?

But maturity also means knowing when AI should not be used. A truly AI-literate person should be able to ask :

Does AI actually add value here ?

Can I verify what it produces ?

Am I exposing sensitive information ?

Does this situation require human empathy, accountability or contextual judgement ?

Would using AI make the task unnecessarily complex ?

And increasingly: Is the value created worth the computational and environmental resources required ?

Responsible AI Literacy therefore also includes digital sobriety. Using AI well does not mean using it everywhere. It means understanding when it is genuinely useful and making that choice consciously.

That kind of judgement cannot be built through a checklist alone. It has to be practised.

 

From a static curriculum to a living learning environment

A curriculum still matters.

It provides structure. It introduces concepts. It can establish a common language around prompting, bias, hallucinations, privacy, security, ethics and responsible use.

But the curriculum cannot be the final architecture.

If AI keeps changing, the learning environment has to change with it.

Schools and organizations need the ability to continuously:

Experiment → Adapt → Measure → Improve

That means being able to edit content as the technology evolves.

• Add new activities.
• Test new models.
• Introduce new organizational policies.
• Update assessments.
• Ground learning in trusted sources.
• And measure whether learners are developing better judgement, not simply whether they completed a module.

This is why AI Literacy increasingly requires an Interactive Learning Platform, not just a content library.

 

AI Literacy should behave like a capability system

A course has a beginning and an end. A capability develops over time.

It is practised, tested, applied, reinforced and updated. Cybersecurity offers a useful comparison: We would never assume that one awareness course permanently makes someone cyber-secure. Threats change. Behaviours need reinforcement. Policies evolve. New situations appear.

AI Literacy is becoming similar. An organization may introduce generative AI today, AI agents tomorrow and increasingly autonomous workflows later.

The capability required at each stage changes. The questions evolve from: What is AI ?

to: Can I trust this answer ?, then: What sources were used ?, then: Should this task be delegated to AI ?, and eventually : How do we govern AI systems that act across multiple workflows ?

A static course library will struggle to support that evolution. A living capability model can.

 

From content creation to experimentation

This shift also changes the role of teachers, instructional designers and L&D teams.

Their value moves away from simply producing learning assets and toward designing environments where learners can investigate, practise, question and demonstrate judgement.

AI can help with that. An AI lesson creator for teachers can accelerate the creation of exercises, scenarios and assessments.

An AI authoring tool for L&D can reduce repetitive production work but faster creation is not the real objective.

The real opportunity is to create richer practice : 

• Instead of showing learners a slide explaining hallucinations, ask them to intentionally trigger one, identify the suspicious claims, verify them and explain what went wrong.
• Instead of defining bias, let them compare responses generated from different contexts and analyze how framing affects the outcome.
• Instead of explaining grounding theoretically, let them compare a generic AI response with one grounded in trusted sources.

That is the shift from content consumption to capability development.

 

Interactive experiences matter because judgement develops through action

AI Literacy is not particularly well suited to passive learning.

People need to manipulate variables, make choices, observe consequences and receive feedback.

This is where an AI-native platform for creating interactive learning experiences becomes particularly valuable.

Teachers and L&D teams should be able to move from : “ I need to create another slide-based module” to: “ I want learners to compare three AI answers, identify which claims require verification, select trusted sources and justify their decision.”

Emerging approaches such as vibe coding for interactive learning can make this much easier:  Instead of being limited to predefined templates or requiring technical development, educators can describe the activity they want and let AI help build the interaction.

This enables interactive learning without technical complexity and gives subject-matter experts much more freedom to turn pedagogical ideas into practice.

The important change is that the tool no longer defines the learning experience. The learning objective does.

 

How Mexty supports a living AI Literacy capability

This is exactly the direction Mexty is designed to support. Rather than treating AI Literacy as a fixed collection of courses, Mexty provides an AI-Native secure Learning Infrastructure where schools and organizations can continuously create, modify, experiment, assess and improve.

The starting point can be an Editable AI-Literacy curriculum :

• Teachers or L&D teams can adapt modules to their audience, policies, sector or maturity level.
• They can update examples as models evolve.
• Add new risks.
• Introduce organizational guidelines.
• Modify activities.
• Create new assessments.
• Reuse trusted institutional content as the source of the learning experience.

But the objective is not simply to make the curriculum editable. It is to make AI Literacy experiential.

 

The AI Literacy Lab

Within the Mexty AI Literacy Lab, learners can explore AI directly in a structured environment: 

• They can compare models.
• Change prompts and context.
• Analyze why outputs differ.
• Test hallucinations.
• Compare general responses with grounded responses.
• Investigate bias.
• Verify information against trusted sources.
• ractise deciding when an AI response should or should not be accepted.

AI stops being only a tool for producing answers. It becomes an object of investigation.

The teacher or trainer remains central: defining the objective, selecting resources, controlling the activity and guiding reflection.

From knowledge to capability

Mexty can combine:

Content → Interactive Activities → AI Experiments → Projects → Assessments → Feedback → Analytics

This allows organizations to move beyond knowledge checks:

• Knowing that AI can hallucinate is one thing but being able to identify an unreliable claim, verify it and explain why it cannot be trusted is a capability.
• Knowing the definition of bias is one thing but being able to recognize it in an output and explain how context or data may have influenced it is a capability.
• Knowing an AI policy is one thing but being able to apply that policy correctly in a new situation is a capability.

That is the difference we need to measure.

 

Security and governance are part of AI Literacy too

AI Literacy cannot be separated from the environment in which learners practise.

If schools and companies want people to develop responsible AI behaviours, the platform itself must model responsible AI use.

That means privacy controls, trusted sources, human validation, controlled access, versioning and clear governance.

A GDPR and EU AI ACT compliant AI learning platform, is therefore not simply an IT requirement. It becomes part of the learning strategy.

You cannot teach responsible AI use inside an environment that does not itself operate responsibly.

 

Measure judgement, not only completion

Completion rates tell us very little about AI capability. Neither does a quiz asking learners to define “hallucination.”

The more meaningful questions are:

• Can the learner recognize an unreliable output ?
• Can they find the missing source ?
• Can they explain why two models disagree ?
• Can they identify privacy risk ?
• Can they determine whether AI should be used for a particular task ?
• Can they justify why human judgement should override an automated recommendation ?

Those are demonstrations of capability and that is where AI Literacy assessment needs to move.

 

From teaching AI Literacy to building AI Literacy capability

The first generation of AI Literacy programs understandably focused on awareness.

What is AI ?
What are the risks ?
How do I prompt ?
What should I avoid ?

That was necessary but it was only the beginning.

The next stage needs to ask :

• Can people investigate ?
• Can they verify ?
• Can they challenge ?
• Can they apply ?
• Can they recognize when AI should not be used ?
• Can they adapt when the technology changes ?

That requires more than a curriculum. It requires a learning environment that continuously supports: Experiment → Adapt → Measure → Improve

This is the shift from teaching AI Literacy to building AI Literacy capability because if AI keeps moving, AI Literacy has to keep moving too and the most important question for schools and organizations is no longer: “Have our people completed the AI Literacy course?”. It is: “Are our people becoming better at understanding, questioning and working responsibly with AI as it evolves?”

That is the capability that will matter.

If you enjoyed this, you’ll love our next articles AI Literacy in France: What Is the French Framework, and How Can Schools Go Further?