The philosophy behind Malta's national AI literacy programme

Last Saturday, we launched AI Għal Kulħadd (AI for All), Malta's national artificial intelligence (AI) literacy programme, starting with the mass adoption course available online at ai4all.gov.mt. 

The course is vendor-agnostic and was developed by the Department of AI at the University of Malta in collaboration with the Malta Digital Innovation Authority (MDIA).  

Anyone in Malta with the e-ID can access it for free and, upon completion, choose a one-year paid subscription, currently offered by OpenAI and Microsoft. 

I am the Principal Investigator of the programme. This article shares the philosophy behind the course because the design decisions that shaped it matter as much as the launch itself. 

We started with the human, not the technology 

When we sat down to design this programme, the first question we asked was not what features to teach or which tools to cover.  

It was a simpler question: What actually makes someone good with AI? The answer turned out to be something most people already know how to do. 

Think about the last time you wrote a message to someone who mattered to you and spent ten minutes getting it right. A friend, a parent, your child.  

In those ten minutes, you were thinking clearly, organising information, and communicating with purpose.  

That is the same skill that makes someone good with AI. It is not about knowing how a neural network works, but the discipline of thinking clearly and knowing what you actually want to say. 

So, that is where the programme starts. From the human in front of the technology, rather than from the technology itself. 

AI literacy is infrastructure 

The programme is not just a course. It is infrastructure, and that distinction shapes every decision we make when building this programme. 

Think about what the road does. It does not tell you where to go, but it makes it possible for everyone to travel further than they could otherwise, to get to work, to see family, to do business.  

A road is a public investment on which everything else builds.  

AI literacy is the same kind of object. The government invests in creating a national baseline.

Every other investment, from the private sector or from individuals, builds on that foundation. The level of the entire country rises with it. 

We studied what other countries have done: Estonia, Finland, the US, Singapore, Lithuania, the UK, and India.  

Each found its own path and flavour that addresses its national needs, just as we are finding ours.
 

We all understood the same thing: you cannot have a digital economy without a digitally literate population, and at this moment, that literacy must include AI. 

The Eurostat numbers tell a clear story 

Three figures from 2025 are worth holding together. Malta is fourth in Europe for personal AI use.  

Maltese workers are first in the European Union (EU) in the use of generative AI at work, with 29.6 per cent reporting use in the last three months of 2025, nearly double the EU average of 15.1 per cent.

And formal AI adoption at the company level in Malta sits at 21.6 per cent, almost exactly the EU average of 20 per cent. 

Read those three numbers together, and a story emerges. Individuals are moving forward at a pace. Organisations are not keeping that pace.  

The gap between enthusiastic individual use, often private and undeclared, and average organisational adoption is both a significant risk and a major opportunity that is not yet being captured. 

That gap is what this programme exists to address.

Not by adding awareness, but by building the shared infrastructure that enables individual capability to translate into organisational capability, and then into national capability. 

This is not a prompting course

We teach prompting, of course, but prompting is downstream of something else.  

A clear thought produces a good prompt, and a good prompt produces useful output from AI. If you want better output, the place to start is not the prompt but the thinking behind it. 

So, we went upstream. Why are you using this tool? What do you actually need from it? 

The challenge with AI is how technology manages information. How we, as citizens, handle information, evaluate it, and decide when to trust it.  

AI is one of the most powerful information tools ever built, making information literacy our first responsibility as a population that uses it. 

We also made a deliberate choice about ethics and safety. Rather than adding a module here and there, we wove them horizontally through everything.  

Wherever you are in the course, you are dealing with the questions around bias, privacy, misinformation, and more. These are not AI problems. They are human problems that AI amplifies, and we treat them that way throughout. 

One message in particular runs consistently throughout the programme. If you are in a personal crisis, find a human. AI is good for almost everything else, but for that, find a human. 

Five dimensions, measured across the whole population

Everything in the programme is calibrated against five dimensions:

1. Theoretical understanding of how AI works
2. Prompt engineering
3. Ethical evaluation
4. Tool integration
5. Strategic oversight, which is to say, knowing when not to use AI.  

That last one is the hardest to teach, and probably the most important. 

Every learner gets an anonymous profile across these five dimensions. That lets us compare across age, gender, region, and level of education.  

Strategically, this matters for two reasons.  

First, it gives us an indication of progress at the population level, which has not been attempted in other national programmes outside Malta.  

Secondly, the aggregated data tells the government where to invest next. If a particular demographic is weak on a particular dimension, that is a finding that colleagues at MDIA can act on directly. 

The measurement instrument itself is new, and we will publish it openly later this year so that any country can use, test, or improve it. 

What keeps it honest over time

There are four things that keep the programme honest as it runs and evolves. 

1. The research team in the Department of AI at the University of Malta leads the science. Evidence-based design, drawn from academic literature, with peer-reviewed publication of findings. This is as much a research programme as it is a delivery programme. 
2. We recruited science communicators and PR specialists to close the gap between academic rigour and public reach. Academic work does not translate itself into something that lands with a retired grandparent or a small business owner. That bridge needs to be built and staffed, and it is. 
3. We have run, and will continue to run, pilots with the public. Real people from different parts of Maltese society shape the content, format, and delivery details before scale. We are designing for the people who actually show up, not for an imagined average citizen. 

4. And we are honest about what the data cannot see. Engagement metrics tell us about completers. They tell us less about people who stopped halfway, and almost nothing about people who never started. Reaching everyone is a distinct design challenge, and we are actively working on it, including the blind spots the data leaves us with. 

AI Għal Kulħadd (AI for All) belongs to everyone who wants to make Malta a better place. 

If you have suggestions that can help us make future actions and deliverables more realistic, we want to hear them. This will make the programme stronger and more relevant. 

The avatars are a feature, not a compromise 

You will notice that the course videos use AI-generated avatars. This was a deliberate decision. 

Firstly, AI avatars can be redeployed in hours. A video filmed with a real person becomes outdated the moment the technology shifts, which in this space could be next week.  

At the national scale, the trade-off between accuracy over time and the comfort of a human face is unacceptable. 

Secondly, a literacy course about AI that uses AI-generated avatars is making a deliberate pedagogical point. People will increasingly encounter this kind of media.  

Evaluating it critically, pausing, asking questions, not simply accepting what you see, is exactly what the course teaches and thus the medium reinforces the message. 

Thirdly, to keep growing, we all need to keep learning on platforms like Coursera, where this course is also delivered.  

AI Għal Kulħadd (AI for All) uses the same platform as the world's leading universities. That familiarity makes it easier for Maltese learners to take other courses on the same platform once they finish this one.  

You own whatever you press send on 

The programme does not judge users on how they use AI.  

Instead, we take an approach that gets the user to reflect on the use: You own whatever you press send on. The responsibility is yours. The judgement is yours. The thinking behind it is yours. 

While this might sound like a warning, it is the good news and the opportunity.

It means AI does not replace us. It instead extends us and gives us more reach, more speed, more capability. However, this only happens if we bring the thinking. 

That is what AI Għal Kulħadd (AI for all) is here to build: A generation of Maltese people who think clearly and use AI well. 

We will be sharing and publishing more about the design decisions, the research, and what we learn as the programme runs.  

The article was originally published on Dylan Seychell's LinkedIn here, and edited. 

----------------------------------- 

Dylan Seychell is a Lecturer in the Department of Artificial Intelligence at the University of Malta and Technical Expert at the Malta Digital Innovation Authority (MDIA). He is also the Principal Investigator of AI Għal Kulħadd (AI for All), Malta's national AI literacy programme.