AI in the Classroom: Innovation or Experimentation Without Consent?

By Courtney King | July 11, 2025

AI is officially in schools.

With little fanfare and even less research backing, generative AI tools are being introduced into classrooms across the United States. In some districts, they’re being framed as digital teaching assistants. In others, they’re handed directly to students with the promise of personalized learning. Just last week, Microsoft and OpenAI announced a multimillion-dollar initiative to train teachers to use AI in the classroom, part of a broader effort to normalize chatbot use in K–12 education.

The stated goal is to help teachers streamline tasks like grading, lesson planning, and writing recommendation letters. Tech companies are also making a case that AI proficiency will be essential for students entering the future workforce.

But behind the slick marketing and workforce-readiness framing, this rollout looks less like educational innovation and more like a mass-scale experiment—one where children, families, and teachers are rarely asked for informed consent.

Old Playbook, New Tech

If this all feels familiar, that’s because it is. In the early 2000s, schools were promised that laptops would revolutionize learning. Districts spent millions, often without clear implementation plans or measurable impact. The narrative then was the same as now: embrace the tech or risk leaving your students behind.

Fast forward twenty years, and we’re watching a near-identical story play out with generative AI. This time, it’s not just devices but intelligent systems—often proprietary, opaque, and untested—being embedded into classroom routines. The pitch is urgent. Students, we're told, will fall behind in a global economy if schools don’t adopt AI now.

The result is a “fear of missing out” model of edtech adoption. Vendors tap into administrators’ anxieties about falling behind on innovation, while also offering discounted student subscriptions around exam periods. That marketing approach is not just aggressive—it’s strategic. Get kids accustomed to AI early, and you secure a pipeline of future customers.

What’s Actually Happening in Schools?

Despite widespread uncertainty about outcomes, adoption is accelerating. In Kelso, Washington, middle and high school students used Google’s Gemini for research and writing assignments. In Newark, New Jersey, Khan Academy’s AI tool grouped elementary students into study cohorts based on skill levels and offered real-time help during lessons.

According to a nationally representative U.S. survey, nearly half of school districts reported providing AI training to teachers as of fall 2024. That’s up from just a quarter the year before—a steep climb that suggests adoption is outpacing both public debate and empirical validation.

Colleges are all in as well. California State University System recently signed a $17 million deal with OpenAI to give over 460,000 students access to ChatGPT. Despite significant budget cuts, the system prioritized AI tools to help students debug code, write essays, and create digital art. Other institutions, like Duke and the University of Maryland, are developing in-house chatbots to support similar use cases.

Do These Tools Actually Help Students Learn?

Here’s the uncomfortable truth: we don’t know. While tech companies promise that AI will personalize learning and close achievement gaps, there’s little real evidence to support those claims. At best, these tools are currently being used as glorified search engines. At worst, they are enabling academic dishonesty and shifting the cognitive load away from students before we’ve studied the long-term consequences.

Julie Kaufman at the RAND Corporation, who tracks national education data, put it bluntly: “It’s really hard to know” whether AI improves learning outcomes. The research just doesn’t exist yet, and the tools are evolving faster than public institutions can keep up.

What we do know is that these systems introduce entirely new dependencies. Students are learning to lean on AI for answers before they’re taught how to assess truth, context, or bias. In many cases, they are surrendering personal data to tools they don’t fully understand, operating within ecosystems where oversight is minimal and commercial incentives are high.

The Bigger Picture: Education Meets AI Governance

This isn’t just about pedagogy or test scores. What’s happening in classrooms right now reflects a broader challenge in AI governance: the gap between innovation and institutional readiness.

Education has always been a battleground for public values. What we teach, how we teach it, and who has access to quality education are deeply political questions. When AI tools are introduced into that mix without strong governance, we risk compounding existing inequities. Under-resourced schools may adopt these tools out of necessity, while more affluent districts have the means to evaluate, modify, or even reject them.

Meanwhile, government support has been inconsistent but increasingly favorable. In the U.S., recent federal messaging has encouraged AI integration at every grade level. The motivation? Maintaining national leadership in the tech race. But that framing treats students less like citizens and more like infrastructure—raw material to be trained for future economic competitiveness.

That’s not education. That’s workforce conditioning masquerading as innovation.

So What Should We Be Asking?

As AI becomes embedded in our education systems, here are a few urgent questions policymakers, educators, and parents should be asking:

• What evidence do we require before a district rolls out AI tools at scale?

• Who audits these systems for bias, accuracy, and privacy compliance?

• How are teachers supported—not just trained—to integrate AI thoughtfully?

• What data is being collected from children, and who owns it?

• Are students being taught to critically engage with AI, or simply use it?

These are not questions for later. They’re questions for now.

The Bottom Line

The debate isn’t whether AI will enter schools—it already has. The real question is whether we have the courage and clarity to shape how it’s used, who benefits, and who gets left behind.

If we want AI in education to be truly transformative, we need more than product demos and early adoption numbers. We need governance frameworks that prioritize student wellbeing, educational equity, and long-term societal impact.

And we need to remember that saying no—or not yet—is sometimes the most responsible form of innovation there is.

Stillwater Insights supports public sector and nonprofit leaders in navigating AI integration with strategy, clarity, and care. We offer advisory services focused on AI governance, compliance readiness, and ethical deployment of emerging technologies. Learn more at stillwaterinsights.io.