AI-Native Education
From AI Tools to AI-Native Learning Systems
Most current approaches treat AI as an add-on. These improve efficiency, but do not fundamentally change how learning works.
AI-native education: learning systems in which AI is embedded in the structure of teaching and learning, where learning processes are co-shaped by human and artificial agents, and gradually evolve toward continuous, adaptive, and collaborative forms.
What Does "AI-Native" Mean?
Defined by system architecture, not features.
AI participates in learning processes
Intelligence distributed across humans & AI
Classroom processes become observable
Systems become interpretable & adaptable
Three Paradigm Shifts
Not isolated features, but interconnected components. A system perspective changes the design question: the unit of analysis is the structure — how components connect, how learning loops form, how AI operates within the architecture.
From local interactions to system-level operation: orchestrating learning processes across individuals and groups, coordinating multiple agents, supporting real-time decision-making.
Visible features sit on top of learning. We focus on system infrastructure — intelligence across layers, teacher-facing decision support, background orchestration.
The Philosophy Behind AI-Native
AI-native education is not the pursuit of a "painless" or "fully automated" classroom. Inspired by Levinas and Han, we are working toward systems that stimulate student agency, respect student alterity, and preserve deep resistance.
On Alterity — Levinas
AI-native education creates an ethical space. We refuse to "totalize" students into data models. We use AI to free teachers so they can return to their infinite responsibility in the face-to-face encounter.
On Depth — Han
We are wary of the "violence of smoothness." A true AI-native system designs productive friction. We offer deep challenges on the path to wisdom.
On Entanglement — posthumanism & distributed cognition
Intelligence is not a plugin — it is infrastructure. Cognition is distributed across students, teachers, and AI agents. This reconfigures pedagogical agency.
The Social Logic of AI-Native — Bourdieu
AI is never neutral. Drawing on Bourdieu's concepts of field, capital, and habitus:
Field — AI as Educational Infrastructure
AI is becoming the field's infrastructure. When algorithms become the underlying architecture, who holds the power to define "good learning"? We build systems like CocoFlow to return that power to teachers.
Capital — From Content Knowledge to Orchestration Capability
A new techno-epistemic capital is emerging — the ability to orchestrate AI agents. Our goal: make this accessible to all teachers and students, not just a technical elite.
Habitus — Guarding Against Algorithmic Habituation
If AI paths are too "smooth," students risk algorithm-dependent habitus. We insist on productive friction — deliberately introducing diverse perspectives and cognitive conflict through tools like CocoNote.
Where We Are Today
✓ Implemented
✓ Early classroom orchestration with AI-powered analytics
✓ Agent-based teaching and learning workflows
✓ Teacher participation in designing AI-supported processes
✓ CocoClass × CocoStudy continuous data loop
○ Open Challenges
○ Large-scale student–AI co-creation across diverse contexts
○ System-level coordination across schools and regions
○ Fully interpretable and ethically transparent learning processes
Beyond Human-Centrism: Toward a Relational View
We care deeply about student-centered learning. But as AI grows more powerful and permeates every layer of our lives, a purely human-centered lens may not be sufficient for understanding the systems we are building and the futures they will shape.
Drawing on Barad's intra-action and Deleuze & Guattari's assemblage, we recognize that students, teachers, AI agents, and learning environments are continuously constituted through their relations. For future practitioners entering an AI-saturated world, the human-machine assemblage itself becomes a critical unit of analysis. Bennett's vibrant matter reminds us that algorithms actively shape possibility. Engeström's activity theory reveals contradictions that drive transformation. Hutchins's distributed cognition shows thinking unfolds across the entire system.
At the same time, this relational perspective must be held in tension with ethical commitments. Levinas insists that each student's alterity must never be reduced to data. Han warns that smoothness erases the friction that makes deep learning possible. Bourdieu demands we ask who benefits and whose trajectories are shaped.
As a system builder, I walk between efficacy and ethics. I am committed to student-centered learning, but I refuse to stop there. As AI reshapes education and society, we must learn to see through a relational lens — understanding human-AI assemblages as they are, not as we wish them to be.
I heed Levinas's counsel on alterity, Han's warning about smoothness, Bourdieu's demand for equity, and the posthumanist insight that agency is always distributed. My research is ultimately about how, in a world saturated by algorithms, we can build systems that are both powerful and humble — systems that enhance collective intelligence while safeguarding what is most precious: responsibility, depth, difference, and respect for the Other.