The same AI tool can support very different kinds of learning

The same AI platform can lead to very different learning experiences. The difference often lies not in the technology itself, but in how educators choose to integrate it into teaching.

In our previous post, we asked what kind of learning AI should support. This raises a natural follow-up question: if we know what we want students to learn, how do we design AI to support it?

One answer lies in something that often receives less attention than the technology itself: the pedagogical framing.

The technology is rarely the deciding factor

Imagine two educators using exactly the same AI platform.

The first introduces it as a place where students can ask questions whenever they get stuck.

The second creates several AI agents with different response styles, each encouraging students to approach the material differently. One agent challenges assumptions, another encourages reflection, while a third gives more direct guidance when students are stuck. We describe this approach in more detail in Use case 1, Engaging students through different response styles.

The technology is identical.

The learning experience is not.

The difference lies in how the technology is framed.

Those choices are not neutral.

They shape how students engage with tasks, what kind of effort is expected, and ultimately what students learn from the experience. This is closely connected to the ideas we explored in Two Schools of Learning in the Age of AI.

Generative AI can be used to check answers quickly or to explore how ideas connect. It can help students complete tasks efficiently, or encourage them to think more carefully about them.

That is what we mean by pedagogical framing.

Researchers do not always agree on how AI should be used in education. But one idea appears again and again:

Technology alone does not determine learning.

How it is integrated into teaching matters just as much.

Different pedagogical framings can lead students to engage with the same AI tool in fundamentally different ways.

Pedagogical framing is about purpose

Pedagogical framing begins before students ask their first question.

It starts with the educator asking:

• What do I want students to learn?
• How should AI contribute to that learning?
• What should remain the students’ own responsibility?

These decisions shape how AI becomes part of the course.

The technology is no longer just available.

It has a pedagogical role.

The educational value of AI depends as much on course design as on the technology itself.

Another example is an educator who designed an AI-supported role-play around pharmaceutical regulation. Instead of positioning AI as a source of answers, students interacted with it as part of a realistic professional scenario where the AI agent performed as the company seeking approval of a medical product, and students were assigned different roles, such as regulator, bioethics expert, or reimbursement specialist.

The technology remained the same, but its educational purpose changed.

We describe this approach in more detail in Use case 3, Using ChatTutor for role-based investigation in a pharmaceutical approval scenario.

From “What can AI do?” to “What should AI support?”

Much of the current discussion about AI asks what the technology can do.

Can it summarise articles?

Can it explain difficult concepts?

Can it generate quiz questions?

These are useful questions, but for teachers, another question may be even more important:

What should AI strengthen in this particular learning situation?

The answer will not be the same in every learning situation.

Sometimes the goal is to help students connect ideas across different topics.

Sometimes it is to encourage reflection.

Sometimes it is to make misconceptions visible.

The same technology can support all of these, but only if it is framed accordingly.

Technology does not create pedagogy

Introducing AI changes the learning environment.

Whether it supports learning depends on how it is designed into teaching.

That includes:

• how students are introduced to AI
• what kinds of questions AI encourages
• what learning goals it supports
• what remains the student’s own responsibility

In other words, pedagogy shapes technology — not the other way around.

Designing AI as part of teaching

This is how we approach the design of ChatTutor: not by asking how AI can do more, but by asking how it can better support different ways of teaching and learning.

That is why ChatTutor is designed to let educators shape how AI participates in learning, rather than simply giving students access to a chatbot.

Instead of asking instructors to adapt their teaching to AI, ChatTutor is designed so AI can be adapted to different teaching goals.

That means thinking about learning objectives before prompts.

Teaching activities before features.

And pedagogy before technology.

If you’re interested in seeing how different ways of framing AI look in real teaching contexts, explore our use cases.

Closing

Ultimately, the educational value of AI depends less on the technology itself than on the learning experiences we intentionally design around it.

The same AI tool can support very different kinds of learning.

The difference lies in the pedagogical choices surrounding it.

The real opportunity is to design learning environments where AI helps students think more deeply, reflect more critically, and learn more intentionally.

Suggested reading

This post draws on established perspectives in learning theory and didactics, including traditions that emphasise learning as a situated, reflective process (e.g. Dewey), distinctions between different approaches to learning (e.g. Marton), and research that connects design, practice, and iterative development in real educational settings (e.g. Brown; Design-Based Research Collective).

Rather than referencing specific works throughout the text, the ideas are reflected in the background and can be explored further through the theories and authors listed below.

Constructive Alignment
John Biggs (1996)
(alignment between learning objectives, teaching activities, and assessment)

Design-Based Research
Ann Brown (1992)
Allan Collins (1992)
Design-Based Research Collective (2003)
(understanding learning through iterative design in real teaching contexts)

Experience and Education
John Dewey (1938)
(learning as experience, reflection, and meaning-making in practice)

Approaches to Learning
Ference Marton & Roger Säljö (1976)
(different ways students engage with learning, not just what they learn)

Constructive Alignment in Higher Education (Contemporary Applications)
Francesca Morselli (2024)
(applying alignment between objectives, activities, and assessment in current higher education contexts)

Critiques of Constructive Alignment
David Newby (2025)
(arguing that alignment frameworks reflect particular assumptions about learning and teaching)