Many schools are ‘using’ or are thinking about using Artificial Intelligence (AI) as a tool. Transformative schools use AI to unlock new levels of thinking, teaching delivery and learning support. The opportunity afforded by AI is not about changing teaching—it’s about helping it work even better for every teacher and learner.
One of the central ideas in @Sandeep Paul Choudary’s book Reshuffle (a highly recommended read) is that sustainable advantage doesn’t come from doing the same thing slightly better; it comes from rethinking the game entirely. In an AI context, this is a maturation of the definition of insanity as we all know it.
In education, AI should move beyond enhancing existing practices. Instead, it should be used as an engine to reimagine teaching, feedback, and unlocking student potential.
In the next two years, schools will fall into one of two categories.
- Schools that use AI as a tool to streamline content, workload and automate admin.
- Schools that use AI as an engine to transform how learning is experienced and how teaching is delivered.
This distinction matters. Schools using AI as a tool to create more content and adjust workflow simply re-digitize old processes. Utilizing AI content capabilities changes the process, not the paradigm. Gemini has recently delivered even more ways for teachers to re-frame content in a purely digital Chrome engagement. Does that really set up the future of learning? We are focused on learning outcomes right?
Devices, dashboards, and digital content are ubiquitous, yet they often overlay a system still bound by a single curriculum, pace, and pathway. Unless we rethink how students learn, receive support, and grow from feedback, are we just accelerating an outdated model?
Schooling still relies on fixed schedules and fixed content. A student’s experience is often shaped by the class they’re in, the teacher they’re assigned, and the curriculum they follow. Progress is measured periodically and support is reactive, usually offered only after a student falls behind.
This “one-size-fits-most” model is the educational equivalent of Facebook’s social graph where progress depends on your context, your connections, your cohort and your ‘position’ in the system.
Some schools are starting to flip that reality, using AI as the engine to help them make learning an ‘all-size’ pursuit. They use AI not just to optimize, but to rebuild learning around real-time student behavior and capability: how they write, think, engage, and respond. Like TikTok's behavior graph, these systems learn from interactions, providing instant, adaptive feedback.
This isn’t just a technical upgrade, it’s also an equity upgrade.
When used as an engine, AI helps begin to level the playing field. Every student, regardless of background, pace, or confidence, receives support that meets them and moves them forward. These systems don't wait for tests or teachers. They build dynamic learning graphs for each learner, responding within their Zone of Proximal Development (ZPD)—the sweet spot for growth.
Traditionally, teachers identify and teach within each student's ZPD through experience and intuition, a difficult task with 30+ students and multiple classes. AI augments this task. AI analyzes student responses, pacing, and confidence, detecting when a learner is ready to be stretched, is stuck, or has mastered a skill. AI adapts and supports real-time feedback, turning every learning moment into a growth opportunity.
Just as TikTok delivers value without pre-set social connections, schools using AI as an engine identify learning needs without waiting for testing cycles or teacher availability. Instead, they build dynamic learning graphs, real-time AI-powered maps of each student's progress and engagement. These AI systems deliver timely feedback and scaffolded support within each student's Zone of Proximal Development (ZPD), where it's most effective.
The ZPD is the ‘learning sweet spot’ where a student can’t quite succeed alone but can thrive with just the right amount of guidance. For example, a student might identify a paragraph's main idea independently but struggle to infer tone or author's intent. A well-timed prompt, question, or nudge can bridge that gap, accelerating learning.
This approach is part of a broader shift toward what we call behaviour-driven learning.
In a behavior-driven model, learning is no longer static. Instead, instruction responds directly to student self-directed learning and engagement. Every pause, answer, revision, or moment of confusion becomes a data point, not for surveillance, but to help the system, and teachers, adjust real-time feedback and engagement.
Traditional learning has been guilty of waiting for students to ask for help or to fail a test before starting an intervention. AI-powered, behavior-driven learning notices when a student misunderstands a concept, rewrites sentences repeatedly, or consistently misapplies a rule, then gently intervenes with feedback and guidance. This is a huge heads-up for teachers looking to respond to each student in their ZPD.
It's not about data collection or monitoring for its own sake. It's about listening to the learner moment by moment, responding with timely and appropriate support.
Together, ZPD-aligned feedback and behavior-driven instruction empower every learner to progress from their current point, not just where the curriculum dictates.
Where we seem to be stuck now is that AI is still promoted as a content / modality generator, auto-marker, or reporting tool. Helpful, yes but it leaves the core paradigm untouched. The timetable still dictates learning; the exam still defines success and the teacher still carries the full cognitive load of tailoring instruction with crazy levels of mind-numbing workload. Teachers around the world are taking time out from teaching with stress and workload being the causal factors.
When AI becomes the engine, schools can scale what great teachers do best: guiding, challenging, and supporting students in the moment they are ready to learn. AI also expands what teaching can be. Teachers move from creators and deliverers of content to facilitators of deeper thinking. They gain visibility into learning they couldn’t otherwise see. They get to spend more time teaching, not just managing and can better integrate parent stakeholding in how to help.
This shift redefines competitive advantage in education, not through selectiveness, branding, or facilities, but by expanding access, increasing responsiveness, and accelerating growth for every learner. For years, school success has been protected by “moats” consisting of reputation, resources or funding. After school tutoring also promotes a similar learning moat. While some insist these moats are too big to cross, time will soon see these moats become creeks, easy to cross in a skip. Just as TikTok bypassed the social graph, AI-powered schools will bypass these traditional moats by making deep, personalised learning available to all at a broader scale.
Eventually, every school will adopt some form of behaviour-driven, ZPD-aware learning. Those who begin to work with AI as an engine, not a bolt-on tool, will define what future-ready education looks like. They won’t just improve outcomes, they’ll unlock access, close learning gaps, and help every student thrive. Teachers will get back to following their passion.
Using AI as a tool can and will enhance your school’s operations.
Using AI as an engine redefines what’s possible for teaching, learning, and equity.