Artificial intelligence is forcing educators to revisit some very old questions. Questions like: What is learning? What is knowledge? What is the purpose of education? At first glance, these might appear to be abstract philosophical questions. Some might even seem beyond debate. Yet they increasingly sit at the centre of contemporary conversations about teaching, assessment, curriculum, and the role of AI in schools.
Spend time reading educational discussions online and you will quickly encounter competing claims. Some educators argue that learning is fundamentally about knowledge stored in long-term memory. Others emphasise inquiry, meaning-making, creativity, agency, or participation. Still others focus on collaboration, networks, and the ability to navigate a world overflowing with information. What strikes me is that these debates often proceed as though everyone shares the same definition of learning.
We do not.
Different educational traditions define learning differently because they begin with different assumptions about knowledge, the learner, the world, and the purpose of education itself. Perhaps before we can have meaningful conversations about AI and learning, we need to spend some time clarifying what we mean by learning in the first place.
Learning as Growth: The Deweyan Tradition
For John Dewey, learning was fundamentally connected to growth. Learning occurred when individuals engaged with meaningful problems, reflected upon experience, and developed greater capacity to participate intelligently in the world.
Knowledge was not something transmitted from teacher to student. Broadly speaking, to Dewey, knowledge was made, not found. It was constructed – and reconstructed – through inquiry, reflection, and action. Dewey relentlessly critiqued a ‘spectator theory of knowledge’. For Dewey, knowledge arises through inquiry, an active, self-regulated process where an individual encounters a problem, forms hypotheses, acts, and reflects on the consequences. He’s best understood as a transactional or pragmatic constructivist. He insisted that learning was an active experience of meaning-making. This remains one of the most powerful ideas in education.
In a Deweyan classroom, students investigate authentic questions. They test ideas. They examine consequences. They connect learning to life beyond school. The focus is not merely on what students know. The deeper concern is who they are becoming. The central question becomes:
How has the learner’s capacity to engage intelligently with the world expanded?
As a history teacher, this resonates deeply with me. The purpose of history education is not simply to remember the past. It is to develop informed, thoughtful, and active citizens capable of participating in democratic life.
Learning as Meaning-Making: The Constructivist Tradition
Dewey differed from later constructivist thinkers who shifted attention towards the learner’s active role in building understanding. To these theorists, students do not passively absorb information. They interpret it. New ideas are filtered through prior knowledge, lived experiences, language, culture, and social interaction.
Within this body of constructivism, however, important differences exist.
Jean Piaget, for example, focused primarily on how individuals construct increasingly sophisticated mental schemas through interaction with their environment. Learning occurs as learners modify existing cognitive structures in response to new experiences. Lev Vygotsky, by contrast, emphasised the social and cultural dimensions of learning. Understanding, according to Vygostky, developed through language, dialogue, collaboration, and participation within communities. Learning is mediated by others before it becomes internalised by the learner.
Despite these differences, both perspectives – like that of Dewey – view learning as an active process of constructing and reconstructing meaning.
This helps explain a reality every teacher encounters.
Two students can sit through exactly the same lesson and leave with very different understandings. Learning is not merely what is presented. Learning is what is made.
Learning as Generation: A Cognitive Bridge
A third perspective focuses on learning as the active generation of meaning.
Generative learning proposes that understanding deepens when learners actively do something with ideas. They explain, elaborate, compare, question, summarise, model, teach others, and create new representations. Importantly, generative learning is not simply about activity for activity’s sake. The central claim is that these forms of mental work help learners to organise, integrate, and connect information with existing knowledge structures. In doing so, they strengthen understanding and support the formation of durable memories.
From this perspective, activities such as explaining an idea in one’s own words, constructing analogies, generating questions, creating diagrams, or teaching a concept to another person are powerful because they promote the cognitive processing that leads to deeper learning.
This is where I increasingly see generative learning acting as a bridge between traditions. To me, it speaks to constructivist notions of meaning-making while also aligning strongly with cognitive science explanations of how durable learning develops.
In my current thinking the key question about learning becomes:
What meaning has the learner actively created, explained, connected, or transformed?
This question feels particularly significant in an AI age. When a machine can generate explanations, summaries, essays, images, and ideas almost instantly, educators must ask whether students are still engaging in the generative work that produces understanding. Some might call this the ‘cognitive friction’ of learning, the challenge of learning…
Learning as Connection: The Connectivist Turn
The emergence of digital technologies prompted further attempts to understand learning in a networked world. Thinkers such as George Siemens and Stephen Downes proposed connectivism as a framework for understanding how learning might operate in environments characterised by information abundance, digital networks, and rapid technological change.
Connectivism remains influential in discussions of digital learning, although it continues to be debated academically. Some scholars view it as a distinct learning theory, while others see it more as a contemporary perspective on learning within networked environments. Regardless of where one stands in that debate, connectivism highlights the important reality that knowledge today increasingly exists across communities, in databases, in organisations, mediated through search engines, shared on digital platforms, and now arguably generated by AI.
In this context, learning therefore developing the capacity to navigate these knowledge ecosystems. In the modern world, students must learn to locate information, evaluate claims, establish connections, and participate responsibly within networks. In such a world, perhaps the central question is:
What new connections can the learner create, navigate, and sustain?
Learning as Memory: The Science of Learning Perspective
Against this background we have a contemporary perspective that might be best described as the ‘science of learning’. This body of research focuses attention on the place of memory within learning. The famous principle popularised by Daniel Willingham captures the essence of this position:
Memory is the residue of thought.
It’s important to note that, contrary to what many might state or imply, the science of learning is not a single theory. Rather, it is a broad interdisciplinary field drawing on cognitive psychology, cognitive science, neuroscience, education, and related disciplines.
Within many contemporary applications of this research, learning is often understood as a relatively enduring change in knowledge or capability reflected through changes in long-term memory. This perspective has had a significant impact on to educational practice.
The body of ‘science of learning’ research has led to understandings in cognitive load, schema formation, retrieval practice, spacing, feedback, modelling, explicit instruction, guidance fading, attention, fluency, and automaticity. Within this field, learning requires attention, cognitive engagement, and the development of durable knowledge structures.
To those embracing this perspective, the central question could be posited as:
What knowledge, understanding, or skill can the learner now retrieve and apply?
To me, this is an important question. But it is not the only important question.
Where the Debate Often Goes Wrong
Educational debates often present these various perspectives on the nature of learning as dichotomous competitors: Knowledge versus inquiry; Memory versus meaning; Explicit instruction versus constructivism; Cognitive science versus progressive education. Yet the longer I teach, the less useful these binaries appear.
The lived experience of the classroom itself refuses to cooperate with them. Certainly, students cannot think critically about ideas they know nothing about. They cannot evaluate historical claims, political rhetoric, misinformation, or AI-generated responses without knowledge. They cannot participate intelligently in inquiry without something to inquire with.
Knowledge matters. Memory matters. Cognitive structure matters.
Yet, at the same time, education cannot be reduced to knowledge storage. Learning is more than the residue of thought.
Importantly, students also need opportunities to question, create, connect, reflect, collaborate, exercise judgement, to explore, and develop agency… to develop character.
The challenge is not choosing one perspective on learning over another. The challenge is understanding what each contributes.
What AI Reveals
To me, artificial intelligence has acted like a spotlight illuminating these differences in perspectives. I’ve been forced to confront my own often unarticulated understandings of learning.
It seems to me, that different definitions of learning lead educators to ask very different questions about AI within the process of learning.
A science of learning perspective might ask:
Does AI strengthen or weaken durable knowledge formation?
A Deweyan perspective might ask:
Does AI support meaningful inquiry and engagement with authentic problems?
A constructivist perspective might ask:
Does AI help learners build, challenge, and revise their understanding?
A generative learning perspective might ask:
Does AI support learner meaning-making, or does it do the generative work for students?
A connectivist perspective might ask:
How does AI reshape the networks through which knowledge is accessed, evaluated, and created?
These are all valuable questions. But they are not the same question.
Towards a Richer Definition of Learning
Perhaps the strongest definition of learning is not found within any single tradition. Perhaps it emerges through synthesis.
Learning requires changes in memory. But it cannot be reduced to memory alone.
Learning involves knowledge. But it is more than knowledge acquisition.
Learning involves activity. But not all activity produces learning.
Learning involves connection. But connection alone does not guarantee understanding.
Learning involves experience. But experience becomes educative only through reflection.
Learning involves generation. But generation must be disciplined by evidence and grounded in knowledge.
A richer contemporary definition might therefore be:
Learning is a durable and meaningful transformation in what individuals and communities know, understand, can do, can question, can create, and can connect. It involves changes in memory and cognition, but also the reconstruction of experience, the construction of meaning, a process of sense-making, the generation of understanding, and participation in wider networks of knowledge and practice.
Perhaps such a definition allows us to honour the insights of cognitive science without reducing education to memory alone. Perhaps it allows us to value inquiry, creativity, collaboration, and connection without losing sight of the cognitive foundations that make deep learning possible.
As teachers, we tend to inhabit the messy middle. I suspect we don’t have the luxury of clinging dogmatically to one narrow conceptualisation of learning.
When we see students wrestling with knowledge and meaning simultaneously… When we watch understanding emerge through explanation, retrieval, reflection, discussion, collaboration, practice, and creation…
Rarely does what we observe and describe as ‘learning’ fit neatly inside a single theory.
Perhaps that is the lesson. The question is not whether learning is memory or meaning.
Perhaps learning is how memory and meaning work together to shape students’ connection to – and participation in – their world.
In a VUCA world increasingly shaped by artificial intelligence, that feels like a conversation worth continuing. Perhaps the challenge before us is not deciding which theory wins. Perhaps it is building a richer understanding of learning that is capable of holding them all in productive tension.
To me, learning involves memory but it is more than memory. Learning is a rich and multifaceted process of memory construction and meaning- making that defies definition into quick ‘hot take’ definitions.
And that may be precisely why, in an AI age, teaching remains such deeply human work.
Disrupted History 
You must be logged in to post a comment.