As AI moves from experimentation into everyday systems and workflows, education leaders must build the capacity, evidence, and governance to use it well.
GUEST COLUMN | by Julia Fallon
For the past few years, much of education’s conversation about artificial intelligence has centered on possibility.
What can AI do? How might educators use it? What tasks could it automate? What new experiences could it create for students?
Those were necessary questions. But they aren’t sufficient anymore.
AI is moving from something educators experiment with to something embedded in the systems, tools, and workflows that shape teaching and learning. Agentic AI will accelerate that shift as systems move beyond generating information toward taking actions and operating across workflows.
That changes the leadership challenge.
The next phase of AI in education isn’t primarily about adoption. It’s about building the institutional capacity to use increasingly powerful technology well.
In September, Eddi, WSIPC, and the Washington School Principals’ Education Foundation (WSPEF) will convene educational leaders in Renton, Washington, for a leadership workshop to explore agentic AI and its implications for K–12 education.
What interests me most isn’t simply exploring what these systems can do. It’s the space being created for leaders to think together before deciding what to do with them: to collaborate, reflect, be curious, test assumptions without immediately turning them into initiatives, and consider where emerging capabilities might solve meaningful problems.
We need more spaces like this.
The speed of AI creates enormous pressure to have answers. But responsible leadership also requires places to ask better questions, test ideas, and learn from one another before emerging practices become institutional defaults.
In many ways, that’s what stewardship looks like at the beginning of technological change.
For me, that conversation starts with four questions.
‘The speed of AI creates enormous pressure to have answers.’
What problem are we trying to solve?
Technology initiatives too often begin with the tool. A new capability emerges, and we immediately start looking for places to use it.
But capability isn’t strategy.
Are we trying to give teachers time back? Help leaders make sense of complex information? Reduce administrative burden? Create better experiences for students?
The goal isn’t to use more AI. The goal is to build better systems for teaching and learning.
What evidence will we collect as we implement?
We often treat evaluation as something that happens after implementation. By then, we’ve purchased the technology, trained people, and built workflows around it.
AI gives us an opportunity to design evidence into implementation from the beginning.
If we’re trying to reduce teacher workload, are teachers actually getting time back? If we’re trying to improve decision-making, do leaders act more effectively? If we’re supporting learning, is the experience improving for students and educators?
Implementation shouldn’t simply produce adoption. It should produce learning about what to expand, modify, or stop.
What guardrails and governance are in place?
Governance is sometimes treated as something that happens after innovation, once people have figured out what technology can do.
That sequence makes less sense as AI becomes more capable.
As systems interact with data, make recommendations, initiate workflows, and potentially take actions, leaders need clarity about where human judgment remains essential, who is accountable, what information systems can access, and how outputs are reviewed.
Good governance doesn’t prevent experimentation. It creates the conditions for trusted practice.
How will we know whether it’s actually improving learning or reducing burden?
Education technology has historically relied on proxies for success: licenses purchased, accounts created, logins recorded, professional learning completed.
Those measures tell us about adoption. They don’t necessarily tell us whether anything became better.
Are students better supported? Are educators better equipped? Are leaders making better decisions? Are systems becoming simpler? Or did the technology simply move the complexity somewhere else?
These aren’t questions designed to slow AI adoption. They’re the questions that allow institutions to move from experimentation to capability.
And they’re not unique to Washington state.
State and district leaders across the country face the same challenge: responding to rapidly changing technology without allowing its speed to determine the pace or purpose of institutional change.
Education has talked about innovation as though introducing something new was itself evidence of progress. AI gives us an opportunity to choose a different measure.
The organizations that navigate this transition best may not be the ones that adopt the most tools or move the fastest. They may be the ones that identify meaningful problems, test new capabilities, learn from evidence, create conditions for responsible use, and let go of what doesn’t improve the system.
That requires something more than an appetite for innovation. It requires the discipline to decide what is worth carrying forward, and the willingness to leave behind what isn’t.
As AI expands what is technologically possible, that may be one of the most important responsibilities education leaders have.
—
Julia Fallon is an education leader with more than 30 years of experience helping states and school systems navigate technological change. She previously served as executive director of SETDA and writes about modernization, institutional capacity, and the leadership needed to turn technological possibility into meaningful outcomes for learners. Connect with Julia at juliafallon.com.
0 Comments