Despite the succession of state laws and school board regulations banning the use of artificial intelligence in classrooms, the AI education market shows no signs of slowing down. For one, it expanded from $150 million in 2021 to $730 million in 2026, and experts estimate it will reach $18.5 billion by 2036.
New technologies almost inevitably follow a hype cycle including massive excitement over the potential, followed by a rapid decline (known among market analysts as the “trough of disillusionment”), before the technology is finally adopted for the right purposes, for the right audiences, and in the right ways. At that point it’s reached the so-called “plateau of productivity,” where best practices have been established and it’s easier to differentiate which products and programs offer the best ROI.
In education, the stakes to get technology right are particularly high, making the hype cycle difficult to navigate. Schools need to be good stewards of tax dollars and savvy spenders of limited budgets amid this initial flood of excitement. But they shouldn’t be so conservative with spending that they fall into the trough of disillusionment and prevent students from accessing tools that will support their academic goals.
WHERE WE ARE NOW
The pressure to get it right is high, and the available information to make the right decision is often low. And that’s exactly where we are today in our AI K-12 journey.
Until now, most of the policy debates surrounding AI use in schools have oscillated between promises to erase equity issues, deliver a personalized learning experience to every student to transform the way we teach and learn, and frantic calls to ban AI in schools entirely. While other countries treat AI like an infrastructure project, injecting it in meaningful ways and developing it like a public good, we have yet to agree what “good AI” looks like for teaching and learning.
Infrastructure, curricular, and learning science play critical roles in supporting district leaders, school leaders, and teachers alike, to make the right choices. Creating deep understanding of what works as AI-powered tools are adopted will help us avoid the frustrating pitfalls of edtech, including mismatched instructional needs, excessive screen time, lack of teacher training, and automation complacency, to name just a few.
To be sure, we’re working to establish the shared infrastructure AI developers need to ensure their tools equip educators with meaningful insight into learning. But harnessing the potential impact of AI in the K-12 space will require a nuanced understanding of how to gauge instructional impact.
THE NEXT STEP
It’s exciting to see some in the field begin taking this next step by thinking through and developing measurement frameworks that connect AI use to the pedagogical practices that research tells us actually move outcomes.
Rather than measuring teacher time saved or platform engagement, for example, what if we could connect AI use to the learning science principles that research tells us drive student outcomes? Brisk Teaching is one such company attempting to do this by mapping each type of AI-enabled interaction to an evidence-based category of instructional practice, then weighting it according to its relative impact on student learning.
With the North Star being the instructional moment, they’re identifying interactions that a teacher facilitates with AI support that wouldn’t have been possible, or as consistent, without it. It then ranks how impactful that interaction was. The goal, of course, is to build a clearinghouse of best practices so district and school leaders have a learning science-grounded view of which AI tools are best suited for their unique needs.
The use of AI in classrooms must foster agency, purpose, curiosity, and connection among students, and help them develop the conceptual proficiency and durable skills learners need to thrive. This is the type of powerful learning AI can support, but the opportunity is not automatic. Emerging technologies simultaneously hold great promise and pose real risks, particularly when used in ways that promote cognitive offloading and decrease the productive struggles that build lasting understanding.
AN URGENT NEED
The work of figuring out how and when AI can advance powerful learning is urgent. While 85% of teachers are already using AI, fewer than half have received any training. Districts manage nearly 3,000 edtech tools on average, but have no shared way to determine which are pedagogically sound, safe, or grounded in evidence of real learning.
The field needs a higher bar for measuring AI’s impact on student learning. Efficiency gains matter, but they are not the most critical measures of a tool’s usefulness or impact. AI solutions providers must be able to demonstrate how their tools create these moments and help districts measure them against their own priorities. This is the rigor the field needs, and the accountability we should demand from every AI provider working in K-12 education.
Jean-Claude Brizard is president and CEO of Digital Promise.
