I have been thinking recently about two ideas that sound quite similar but, I believe, point to something very important in education: learner agency and epistemic agency.
We often talk about wanting students to become independent learners. We want them to set goals, make choices, manage their effort, find the resources they need, decide how to approach a problem and take responsibility for their progress.
This is learner agency. But there is another layer that interests me even more: epistemic agency.
It is one thing for a student to take responsibility for learning. It is another for them to take responsibility for knowing.
So, what is the difference?
Imagine a student working on a difficult problem.
A student exercising learner agency might ask:
What am I trying to achieve? Which approach should I use? Where can I find help? Should I try another strategy?
These are important questions. The learner is no longer simply following instructions; they are actively directing their learning.
But an epistemically agentic student begins asking another set of questions:
How do I know this is true?
What evidence supports it?
Can I trust that evidence?
Is this explanation actually convincing?
Could there be another explanation?
What would make me change my mind?
That difference matters.
Learner agency is about having greater ownership of the learning process. Epistemic agency is about having greater ownership of knowledge itself—how we construct it, evaluate it, question it and revise it.
And then AI arrived...
This distinction has become much more important with Generative AI.
Today, a student can ask AI to explain a concept, solve a mathematical problem, summarise a paper, construct an argument, write code, suggest evidence or produce a polished response in seconds.
That is an extraordinary capability.
But it creates a slightly uncomfortable question for educators:
Who is actually doing the thinking?
A student can now produce more while potentially thinking less.
The final piece of work may look excellent. It may be well structured, sophisticated and technically impressive. But that tells us surprisingly little about what happened intellectually between the student and the AI.
Did the student make the important decisions?
Did they challenge the answer?
Did they check the evidence?
Did they notice an assumption?
Did they compare alternatives?
Did they decide what should ultimately be accepted or rejected?
Or did they simply accept what the AI produced?
For me, this is where epistemic agency becomes particularly important in AI-supported learning.
The question is not simply whether students use AI
I increasingly think that asking whether students should or should not use AI is too narrow a question.
A more interesting question is:
What functions are we giving to AI, and what functions are we deliberately keeping with the learner?
There is nothing inherently problematic about AI generating possibilities, suggesting approaches, providing feedback or offering alternative explanations.
In fact, these capabilities can create very powerful learning opportunities.
Suppose a student asks AI to generate three explanations for the same phenomenon. The student then compares them, investigates the evidence behind each one, identifies weaknesses, challenges assumptions and finally constructs and defends their own position.
AI has done a great deal of work.
But so has the learner.
More importantly, the learner has retained the judgement.
That is very different from asking AI for the answer and accepting whatever appears on the screen.
So perhaps the important distinction is not simply between AI use and non-AI use.
It is between AI use that strengthens human agency and AI use that gradually transfers that agency away from the learner.
We need more than AI literacy
We understandably talk a great deal about AI literacy: knowing how AI works, understanding its limitations and learning how to use it effectively.
All of that matters.
But I think we need something more.
We need students who can exercise judgement in the presence of AI.
Students need to be comfortable saying:
This answer looks convincing, but I need to verify it.
The evidence does not support that conclusion.
There is an assumption here that I want to challenge.
AI has given me a useful starting point, but I disagree with its reasoning.
Or simply:
I don't know yet. I need to investigate further.
That final sentence may actually be one of the most important signs of genuine learning.
From learning to knowing
Perhaps this gives us a useful way of thinking about education in the AI era.
Learner agency asks:
Am I taking responsibility for how I learn?
Epistemic agency asks:
Am I taking responsibility for what I accept as knowledge—and why I accept it?
We need both.
Because the future of education cannot simply be about giving students faster access to answers. Answers are rapidly becoming abundant.
What will remain enormously valuable is the human capacity to question, evaluate, interpret, challenge, connect, revise and ultimately judge those answers.
That is why I increasingly see the relationship between learner agency and epistemic agency as central to education in the age of AI.
Learner agency gives students ownership of their learning.
Epistemic agency gives them ownership of their knowing.
And perhaps our most important responsibility as educators is to make sure that, as AI becomes more capable, our students do not become less so.
© 2026 [Dr Theo Mojtaba Ammari] All Rights Reserved.




