Who Builds Learning Now? AI and the Democratisation of Education

AI has made it possible for almost anyone to create something that looks like learning.

A learner can upload a document and ask for a summary. A manager can turn a PDF into a quiz. A subject matter expert can ask for a course outline, learning objectives and an assessment in a few minutes.

That is a real shift. I do not think we should minimise its importance. For the first time, people can create useful learning materials without always needing a platform, a production team or an instructional designer.

But we need to be careful about what we mean by learning.

There is a difference between discovering the answer to a question and building knowledge. And there is a difference between a useful learning episode and a serious learning journey.

AI is already very good at the first part. If you want to solve a small problem,  follow a simple recipe or deal with a very specific context, AI can do a great job. It can give you a quick answer, explain a concept, create a quiz, summarise a document or help someone understand something in the moment.

A simple example. I can upload two chapters from a history book and create a questionnaire for my daughter. I do not need anything else. I can build something like that in a moment and we can discover what she does and doesn’t know.

That is useful, and in some cases, it is enough.

But helping my daughter learn history by interacting directly with AI is a very different thing.

That distinction matters because much of the conversation around AI and learning is treating those two things as the same, which they are not. A learning episode can be instant, useful, and personalised. A learning journey is something else entirely. It has structure and direction. It asks the learner to move from not knowing, to knowing, to applying, to reflecting, to ultimately changing the way they think or act. That is much harder.

Access to creation is not mastery of design

When people talk about AI democratising education, I think this is where the argument becomes interesting. AI has absolutely democratised access to creation. More people can now create learning materials. More subject matter experts can turn their knowledge into something structured and more learners can get help at the exact moment they need it.

But access to learning creation is not the same as mastery of learning design.

If someone wants to learn something simple, AI may be enough. But if someone wants to learn nuclear physics, leadership, clinical judgement or any serious capability, the person learning has to do much more than take  a course. They have to take control of the learning journey. They need to understand whether the content is good, whether the next step makes sense, and whether they are really assimilating knowledge.

That requires them to use all their metacognitive skills to direct the learning. In reality, most people do not want that burden. They do not want to spend their energy constantly deciding whether the learning process  is correct. They just want to learn.

This is the first big limitation of prompt-only learning. It looks empowering, but it asks a lot of the individual. They have to choose the right input, ask the right question, judge the quality of the answer, redirect the system when it goes wrong, and all the while keep themselves motivated.

AI-assisted learning is a continuum

I do not think the future is simply “AI creates the course”. The way to think about AI-assisted learning creation is as a continuum. At one end, there is prompt-only creation. A user asks AI to create a beginner course on a topic. It is fast and useful as a starting point. But it is also limited. It may sound coherent and polished, but that does not mean it has strong pedagogy behind it or strong alignment to the learner’s real goal.

Then there is resource-directed creation. Here, AI works from real source material such as a video, a manual, a lecture, an article. That is stronger because the system is grounded in something factual.

Next comes goal-directed creation, where the starting point is not just content, but an outcome. What capability are we trying to build? What should someone be able to do afterwards? How will we know they can do it?

Another route is journey-directed creation. Here, the educator or instructional designer shapes the experience. Where does the learner begin? Where should they struggle? When should they reflect? When should they practise? When should they be assessed?

Finally, there is human-curated creation, where the expert remains closely involved in the structure, sequencing, assessment and meaning of the experience. AI is still there, but it is not the director. It is the assistant.

This distinction is important because learning is not one thing. It depends on how you define learning and what kind of learning you are trying to create.

When we are talking about critical thinking, deep learning, long journeys, big goals and important decisions, we need human-directed learning.

The same is true of the assets we give AI. There is a big difference between asking AI to create something from a blank prompt and providing it with a rich set of learning assets such as source materials, activities, learning paths, assessments and full programmes. The more serious the learning goal, the more the quality of those inputs matters.

The issue was never only content.

This is where I think many organisations will make mistakes. They will see that AI can produce more content, more quickly, and they will assume the problem is solved.

In fact, every time there is a technological revolution, we oversimplify learning. When radio arrived, people said there was no reason to go to school because students could listen. When cinema arrived, people asked why we still needed books. Now AI has arrived, and we risk doing the same thing again. We risk assuming that because a new medium can present information, it can replace the deeper act of learning.

The biggest problem is the simplification of what learning is.

Learning is not just information transfer. It is not only cognitive. It is social. It is emotional. It is cultural. It carries values. Learning comes with values and learning comes with visions of the future.

That is the difference between functional learning and transformative learning. Functional learning might show someone how to make a cup of coffee by following a recipe.  Use this amount of water, this amount of coffee, this method, this timing. Transformative learning helps them understand how to make really good coffee.  How grind size changes flavour, how water temperature affects extraction, how taste develops, how to adjust when something is wrong, and eventually how to create something of their own. That sounds like a small distinction, but it is not. It changes what we think the learning is for.

There is a useful way to think about this. Learning has a politics and an art.

That does not mean politics in a narrow party-political sense. It means learning is never neutral. We always make decisions about what matters, what should be challenged and what kind of future we are pointing towards. Learning is always political in that broader sense, because you make a decision on the road you are taking. We always put emotion, values and perspective into it.

And learning is also art, because it is about how people sense, understand and interpret the world while the world around them is changing. AI can help with this, but it cannot own it.

This is where the educator and instructional designer become more important, not less. As AI takes on more of the actual production work, their value shifts towards shaping the purpose, structure and direction of the learning experience.

In the past, a lot of instructional design time was spent producing the pieces of learning. At LearnWorlds we’re seeing this shift directly: instructional designers are no longer having to spend so much time creating the content, the image, the video or the asset. They can spend much more time on the details and refinements of the learning journey.

That is the important shift. The educator moves upstream. They become less of a content producer and more of a director.

Take any great novel that has been adapted more than once. The source material may be the same, but two directors will not make the same film. The tone, the pace, the atmosphere, what is emphasised and what the story seems to mean all come from the person directing it, because they bring  their own personal interpretation and perspective.

Learning works like that too. Learning is a storytelling experience, almost a cinematic experience. The same source material can become many different learning experiences. Learning is not one story from the same book. There are many different stories.

That richness matters. It is not an inefficiency to be removed. It is part of what makes learning human and it is part of what gives us different perspectives, ways of thinking and culture.

This is where I think AI becomes exciting, rather than threatening. It gives educators a larger design space. This is why we talk about directors and designers. Educators can now open the design space, create prototypes of a learning journey, explore different possibilities, and go deeper into the design than they could before.

AI can help generate the material, but the human still decides what the experience is trying to do. That is the essential difference between content generation and learning design.

Direction turns material into learning

Without that direction, there is a danger that organisations will treat an AI interaction as if learning has happened. AI gives you a perspective of what exists. It is just a perspective. Sometimes it is a very good perspective. Sometimes it is surprising and useful. But a unique dialogue or a couple of interactions should not be treated as if the learning is finished.

There is another risk too. When a learner is alone with AI, something social can be lost. Learning is social. Historically, ideas, disciplines and movements have developed through people, communities, disagreements and shared standards. When it is just the learner and the AI, we lose some of that.  Private dialogue with AI can remove some of the social components that help people, and societies, test and advance ideas.

AI can personalise the process to an extraordinary degree, but millions of separate conversations between individuals and machines are not the same as a shared intellectual culture. If learning becomes too isolated, we risk losing some of the disagreement, negotiation and collective sense-making through which knowledge advances.

The question is not only whether an individual can learn effectively with AI. It is what happens to collective human progress when so much learning takes place in private.

That does not mean we should reject AI. Quite the opposite. I am the first and biggest ambassador of this part of the technology. AI can give us just-in-time support. It can understand context. It can work with documents. It can create useful answers very quickly.

But we need to understand what kind of learning we are asking it to support.

For organisations, this becomes a practical question. Where is AI enough? Where do we need directed learning experiences? Where do we need assessment, feedback, validation and a human point of view? Where do we need people to become better users of AI itself?

Because if we expect people to work with very intelligent copilots, they also need to become more skilled. If you are working with a very intelligent system, you have to raise your game. You have to understand why it is saying what it is saying.

So yes, AI has changed who can build learning. It has opened creation to more people. It has made many kinds of learning faster, easier and more accessible.

But the more AI democratises creation, the more we need to understand direction. The politics and art behind learning.

The future will not be built by pretending AI can do everything alone. It will be built by people who understand when AI should generate, when it should assist, when it should adapt, and when a human being still needs to direct the experience.

Prompting can create material, but direction is what turns that material into learning.