The artificial intelligence in teaching market is experiencing explosive growth, projected to nearly triple from $1.71 billion in 2025 to $5.55 billion by 2030, but a critical gap is emerging between the tools schools purchase and whether students actually use them. A year-long study of AI deployment in real classrooms reveals that implementation infrastructure, teacher support, and school leadership determine success far more than the sophistication of the technology itself.

The market expansion reflects genuine demand. A U.S. Department of Education report showed that generative AI adoption among primary and secondary educators jumped from 17% in April 2023 to 42% by November 2023, signaling rapid institutional acceptance. Major technology companies including Google, Microsoft, IBM, and OpenAI are all competing for market share, while specialized edtech firms like Carnegie Learning are introducing new tools such as LiveHint AI, a generative AI math tutor that provides personalized feedback and problem-solving pathways.

Yet the real story emerging from schools using these tools tells a different narrative. Researchers tracking 10 grantees deploying AI-powered learning tools across K-12 classrooms found dramatic variation in actual usage, even when schools were using identical products. One literacy tool achieved 98.5% registration rates in a North Carolina district but saw only 28% of students actually using it, while the same tool reached 52% utilization in a large urban Georgia district. A math tutoring program showed engaged usage ranging from 0.7 to 5.6 days per week across different schools in the same charter network.

What Separates High-Impact AI Classrooms From Low-Impact Ones?

Schools that successfully deployed AI tools and met their usage targets shared specific characteristics that had nothing to do with the technology itself. These schools built the tool into the regular school day rather than treating it as an optional add-on, assigned a site-level adult to actively support participating teachers, and planned for predictable disruptions like weather days, testing windows, and staffing gaps. This mirrors the implementation infrastructure that has distinguished high-impact tutoring programs over the past four years.

Teachers also exercise significant discretion over how AI tools get used once they arrive in classrooms. In one Arizona district, a math tool purchased specifically for intervention with struggling students was instead used most heavily for general instruction, contradicting the district's original guidance. This suggests that procurement decisions alone cannot determine outcomes; teacher agency and classroom context shape whether AI tools amplify learning or become expensive distractions.

How to Build AI Implementation That Actually Works

• Integrate into regular instruction: Schools should build AI tools into the standard school day rather than positioning them as optional supplements, ensuring consistent student exposure and engagement.
• Assign dedicated support staff: Designate a site-level adult to actively support teachers using the tools, helping troubleshoot technical issues and model effective implementation practices.
• Plan for operational disruptions: Account for weather closures, testing windows, staff turnover, and other predictable interruptions that can derail usage patterns and engagement metrics.
• Connect usage data to academic outcomes: Invest in interoperability standards so AI platform engagement data flows into the same student information systems where academic progress is tracked, enabling leaders to assess whether tools are actually working.
• Establish accountability frameworks: Set clear expectations about when tools will be used, who monitors adoption, and what happens when engagement falls below meaningful thresholds, potentially including outcomes-based contracting.

The research also highlights a critical infrastructure gap. Many districts are purchasing AI tools without the ability to measure whether they're producing learning gains. Data-sharing agreements, institutional review board timelines, and partner cooperation challenges have proven to be the hardest obstacles for researchers trying to evaluate tool effectiveness. States are best positioned to address this by investing in streamlined data infrastructure, model data sharing agreements, and research-friendly procurement policies.

What Does the Evidence Actually Show About AI Learning Gains?

Despite the market's rapid expansion, the field still lacks clear evidence about which AI tools produce learning gains for which students under which conditions. Researchers expect outcome results from the year-long study to arrive later in 2026, but as of now, the critical question remains unanswered: Which tools actually help kids learn more ? This absence of clear evidence creates a risk that edtech becomes an inefficient use of student time and fuels broader backlash against technology in schools.

Experts emphasize that AI tools should amplify human instruction, not replace it. Speaking at ISTELive 26, education leaders stressed that effective learning requires balancing digital tools with foundational skills like handwriting and mathematical reasoning. Research shows that cursive handwriting produces more widespread brain connectivity than typing, promoting deeper cognitive processing and better recall. Similarly, students who develop strong number sense and geometry understanding before age six show long-term academic success, suggesting that digital tools should come after conceptual foundations are secure, not as shortcuts to them.

"Teachers need to remain at the center of the education. You're the professionals, you know your students, you're with them every day. You are the experts. This is what tech can't do, and this is what we are constantly saying about AI. You're supporting the student's emotional intelligence, and you can't ignore classroom culture," said Kathleen Ouellette, chief education and innovation officer at VictoryXR.

Kathleen Ouellette, Chief Education and Innovation Officer at VictoryXR

As more than 130 AI-in-education bills move through state legislatures across 31 states, and at least 28 states have published official AI guidance for schools, policymakers face a critical choice. They can focus on which tools to allow or restrict, or they can invest in the infrastructure that makes evidence possible. The research suggests the latter approach will ultimately matter more for student outcomes than any specific tool selection.

The AI education market will continue its rapid growth, but schools and districts that succeed will be those that treat implementation as seriously as procurement, support teachers as actively as they support technology, and remain focused on whether students are actually learning, not just whether they're using the tools.