On Learning, AI, and System Design

My early work was deeply shaped by constructivist and constructionist perspectives, particularly those of David H. Jonassen and Seymour Papert.

Subsequent engagement with The Cambridge Handbook of the Learning Sciences and The International Handbook of the Learning Sciences expanded this into metacognition, self-regulated learning, knowledge building, orchestration, scripting, project-based learning, evidence-centered design, and design-based research.

This transition reflects a deeper shift: from designing tools to designing learning systems.

I do not treat product development and research as separate activities.

Real classroom systems are sites of intervention, but also sites of inquiry. They make it possible to ask: How does AI mediate teaching and learning? How do teachers engage in AI-supported design? How do learning processes unfold in real classrooms?

Practice is not downstream from theory. It is one of the sites where theory is generated, challenged, and refined. This position aligns with design-based research, but is grounded in real educational constraints.

Much of the current discourse on AI in education remains instrumental. AI is framed as a tool for efficiency, automation, personalization, and productivity.

These framings are theoretically limited. They assume that AI operates on top of stable educational structures.

This reframes the unit of analysis:

AI should not be understood as a feature layer added to existing tools. Instead, AI becomes part of the architecture of the learning system — including orchestration of learning processes, coordination of multiple agents, mediation of interaction and feedback, and integration of data, action, and interpretation.

One of my central concerns is agency in educational systems.

Teachers should be understood as designers — and increasingly creators — of AI-mediated learning environments.

AI does not simply extend human capability; it reorganizes human agency within teaching systems. It transforms pedagogical agency: how teachers define problems, design learning processes, interpret student activity, and exercise judgment in context.

This shift is not only technical, but epistemic: it changes how teaching is conceived, enacted, and studied.

Concepts such as "human-in-the-loop" and "human–AI collaboration" are useful, but they often remain within a fundamentally human-centered ontology.

I draw on posthumanist perspectives to explore learning as emerging from relational assemblages. In this view: agency is distributed across human and non-human actors, cognition is extended beyond the individual, and learning is situated within socio-technical systems.

The question is not whether humans remain central, but how different forms of agency are reorganized:

The following perspectives are used as conceptual tools to understand how learning systems are constituted, how they operate, and what tensions they produce.