Designing Teacher Development Systems: Day 2 Notes

From a working group on education policy in Sri Lanka

Day 2 felt like a turning point.

If the first session was about understanding the problem at a high level, this was where the work began to take shape — not by moving faster, but by going deeper. The day unfolded in two distinct parts. The first expanded the conversation outward, forcing us to think about how education itself is changing. The second brought us back into the system, asking what we could actually do within it. What made the day interesting was not that these were separate conversations. It was that they were clearly connected.

AI in Education: Rethinking the Foundations

The AI seminar on Day 2 did not begin with tools, or classrooms, or even education.

It began with history.

The discussion was framed through the lens of the industrial revolutions — not as a way of adding context, but as a way of setting scale. Steam power reshaped economies. Electricity reshaped society. Computing reshaped how humans interact with information. Each of these moments did not simply introduce new capabilities; they changed what it meant to function in the world.

AI, we were told, needs to be understood in exactly the same way.

Not as an innovation within the system, but as something that sits outside it and forces the system to change.

Once that framing was established, the conversation moved into defining AI itself — and here, one of the most useful corrections was made early.

AI is not ChatGPT.

It is not even generative AI.

It is a layered ecosystem that spans from basic statistical reasoning all the way up to complex neural networks and large language models. The now-popular tools that dominate public discussion sit at the very top of this stack, but they are built on decades of developments in mathematics, computing, and data systems.

This matters because it reframes the conversation from hype to structure.

If AI is treated as a single tool, then the response is to train teachers to use that tool. But if AI is understood as a foundational layer of technology — like computing itself — then the question shifts from usage to adaptation.

The system is not adopting a tool. It is adapting to a new environment.

This is where the seminar began to move from explanation to implication.

Because once you understand that AI is likely to become as embedded as computing — something students will inevitably use in their future work — then the role of education becomes less about exposure and more about preparation.

And that preparation cannot begin with students.

It has to begin with teachers.

At this point, the conversation introduced a framework that made the discussion more concrete: the four pillars of AI adoption in education.

Rather than focusing narrowly on classroom use, the model expanded the scope of what “adoption” actually means. It suggested that integrating AI into education is not a single intervention, but a system-wide shift that requires four things to happen simultaneously: building basic AI literacy, creating pathways into AI-related industries, supporting teachers in adapting their practice, and ensuring that administrative systems are capable of enabling and sustaining these changes.

What is useful about this framing is that it immediately exposes the limitations of how AI is often discussed.

Most conversations focus on the first pillar — literacy — and occasionally touch on student pathways. But teacher support and administrative readiness are treated as secondary concerns, when in reality they are what determine whether anything actually changes.

Without them, adoption remains superficial.

This became even clearer when the seminar introduced the Gartner Hype Cycle.

The model itself is simple: new technologies emerge, expectations rise rapidly, disillusionment follows, and eventually, if the technology survives, it stabilizes into practical use.

What made this relevant was where AI currently sits on that curve.

Generative AI, particularly large language models, are positioned near the peak of inflated expectations — a point where possibilities appear limitless, but practical understanding is still immature.

This has two immediate implications for education.

The first is that much of what is being said about AI right now is likely to be overstated. The second is that a period of disappointment is almost inevitable.

And yet, this does not make the technology less important. If anything, it makes it more important to think carefully about how it is integrated.

Because systems that respond only to the hype tend to over-invest early and withdraw too quickly when results do not immediately follow.

Education systems, which operate on much longer timelines, cannot afford to behave that way.

The seminar then shifted into something more practical.

Instead of asking abstract questions about AI, participants were asked to explore how generative AI could actually be used within education. This is where the EdTech Insiders resource was introduced — not as a list of tools, but as a map of possible applications across the education system.

What this exercise did was expand the imagination.

AI was no longer just something that generates text. It became something that could: support lesson planning, assist with feedback and assessment, enable personalized learning experiences, and even influence administrative workflows.

But at the same time, it revealed something more subtle. The value of AI is not in any single use case. It is in how these use cases connect.

Because when you look at these applications together, a pattern begins to emerge. AI consistently reduces the cost of producing outputs — whether those outputs are lesson plans, explanations, or assessments. And this creates a problem. If outputs become easier to generate, then outputs themselves lose value. This applies to students, but it also applies to teachers.

This is where the seminar reached its most important insight, even if it was not stated directly. AI does not just change what we can do. It changes what matters. If content can be generated instantly, then teaching cannot be defined by content delivery. If answers can be produced on demand, then learning cannot be measured by answers alone. And if both of these are true, then teacher development cannot be about training teachers to do what they already do more efficiently. It has to be about helping them do something different.

This is where the constraints discussed in the seminar become critical.

In a context like Sri Lanka, AI adoption is not just limited by awareness. It is constrained by infrastructure, uneven access to technology, limited training, resistance to change, and broader economic realities.

These are not minor barriers. They fundamentally shape what is possible.

Which means that any serious attempt to integrate AI into education must begin with the system as it exists, not as it is imagined.

By the end of the seminar, the conversation had moved far beyond AI as a topic. It had become a way of interrogating the system.

It exposed how fragile some of our assumptions about teaching and learning are. It highlighted how unprepared existing structures are for rapid change. And it made it clear that the challenge is not technological, but institutional.

The most useful takeaway from the session is not a specific recommendation.

It is a shift in perspective. AI is not something that needs to be inserted into education. It is something that forces education to redefine itself. And any attempt to design teacher training and professional development going forward will have to start from that point. Not with tools.Not with platforms. But with a much more difficult question: What does it mean to teach, in a world where intelligence is no longer exclusively human?

From Policy to Practice: Working With the System

If the seminar expanded the conversation outward, the working session brought it back into focus.

The task was to begin engaging directly with the three foundational policy documents — NEPF 2022, NEPF 2023, and TGES 2025 — and to identify what within them relates to teacher training and professional development.

At first, this seemed like a process of extraction. But it quickly became something else.

Teacher development does not exist in these documents as a single, clearly defined system.

It appears in fragments. It is embedded in discussions on human resources, in references to teaching and learning, in policy statements on standards and accountability, and in broader reform agendas. It has to be assembled from across the text.

And in the process of assembling it, the gaps begin to appear.

One of the earliest and most important decisions the group made was to narrow the scope of the work.

There was an initial instinct to look at teacher development across all levels of education — from early childhood through to tertiary and vocational systems. But it became clear that this would lead to a level of abstraction that would make meaningful design difficult.

So the focus shifted. The group chose to concentrate specifically on teacher training and professional development within general education. This was not a reduction in ambition. It was a recognition that depth requires focus. Each sector of education has its own complexities, and attempting to address all of them simultaneously risks producing something that works for none of them in practice.

With that decision made, the nature of the exercise changed. The group was no longer mapping the entire system. It was trying to understand a specific part of it well enough to design within it. And that required a different way of reading policy.

The documents themselves offer clear signals about direction.

There is a push toward professionalizing teaching through structures such as a national teacher council, formal standards, and licensing mechanisms. There is recognition of the need for continuous professional development, improved working conditions, and clearer career progression pathways. There is also a broader shift toward learner-centered pedagogy, competency-based assessment, and integration of new forms of learning.

Taken individually, these ideas are coherent. But reading them together raises a more difficult question. Do they form a system that can actually function?

This is where the working group began to move beyond interpretation.

The goal was not to summarize what the documents say, but to understand what they require.

Every policy statement carries an assumption about how it will be implemented. If teachers are expected to adopt new pedagogical approaches, then there must be mechanisms through which they learn those approaches. If professional standards are introduced, then there must be systems to enforce, support, and evaluate them. If continuous development is emphasized, then it must be structured in a way that is actually continuous in practice.

And it is at this level that the gaps begin to emerge.

One of the most consistent themes in the discussion was the absence of continuity. There is strong recognition across the documents that teacher development must extend beyond initial training. But the structures that make that development ongoing are not clearly defined. Without clarity on how, when, and by whom this development happens, it risks remaining an aspiration rather than becoming a system.

Another tension emerged around alignment. Curriculum reform is moving toward more complex, skill-oriented forms of learning. But teacher development does not always move at the same pace. Teachers are expected to deliver new forms of education without necessarily being supported in learning how to do so themselves. This creates a structural mismatch.

Teachers become implementers of change, rather than participants in it.

This is where the earlier discussion on AI becomes relevant again.

Because it highlights the same issue from a different angle.

The system is asking teachers to operate in increasingly complex environments — with new pedagogies, new expectations, and new technologies — but the structures that support their development have not fully caught up.

By the end of the session, there were still no answers. But there was clarity. The scope had been defined. The documents had been engaged with critically. And the work had shifted from understanding policy to working with it.

If Day 1 was about recognizing that something is missing, and the seminar in Day 2 was about understanding how the world is changing, the working session revealed something more grounded.

The challenge is not a lack of ideas.

It is the difficulty of turning those ideas into something coherent, connected, and implementable within the realities of the system.