Artificial intelligence has quietly entered the nation’s classrooms. Teachers and administrators are scrambling to catch up.

On April 23, the White House announced the Advancing Artificial Intelligence Education for American Youth Executive Order, creating a federal taskforce chaired by the White House Office of Science and Technology Policy, the Department of Education, and other agencies. The goal: develop strategies for teacher training, student competitions, and targeted AI project funding.

That same day, Bojan Tunguz, one of the authors of this piece, published a report through the American Enterprise Institute (AEI) that closely aligns with the White House initiative. But the report argues AI isn’t just another disruption to manage with new courses or budget lines. The latest models rival graduate-level reasoning, upending how students learn, how teachers teach, and how society certifies knowledge. We need more than coordination—we need to reinvent educational practice.

For the first time in history, expert-level knowledge is free and instant. The internet democratized access to information, but direct interaction with subject-matter experts was still reserved for elite institutions. Now, any learner can ask a complex question at midnight and receive a coherent, often accurate, answer seconds later.

This expertise comes with a powerful feedback loop. Instead of waiting days or weeks for a graded paper, students can revise essays, solve problems, or debug code in real time. This closes “learning latencies”—the lag between teaching and feedback—that have long hindered student progress. Research shows that timely feedback accelerates skill acquisition; AI takes that acceleration to another level.

AI also makes personalized instruction scalable. A student struggling with quadratic equations can get tailored explanations, while a peer ready for more can explore orbital mechanics. Differentiated instruction was once a dream—now it can be routine.

Additionally, AI unlocks “just-in-time” education. Traditional curricula teach broad knowledge “just in case” students need it later. But workplace skills evolve faster than syllabi. With AI, learners can quickly pick up specific skills needed for tomorrow’s project and move on. Foundational knowledge still matters—students must recognize when AI outputs nonsense—but the days of multiyear runways to skill application may be numbered.

Unfortunately, AI’s most visible impact in classrooms so far is cheating. Students ask ChatGPT to write essays or complete assignments. Teachers respond with detection tools or fall back on in-class blue-book exams.

But cheating is a symptom, not the disease. The real problem is that our tests were built for a pre-AI world—focused on recall and routine tasks that AI now automates. Instead of patching this system, we need assessments that measure what still sets humans apart: framing novel problems, integrating ideas, scrutinizing sources, and exercising judgment about when and how to trust algorithms. In many cases, effective use of AI should itself be a graded skill—much like citation management or statistical software.

Credentials also must evolve. Degrees and standardized certificates are one-size-fits-all and change slowly, even as industry needs shift rapidly. If businesses require a new data analysis method this year, a degree program updated next decade is too late. Micro-credentials and dynamic skill portfolios, continuously validated by AI-driven assessments, offer a faster, more precise alternative. Employers could focus on demonstrated competencies, not outdated transcript entries.

Implementing these reforms will require action at every level. The White House is right to focus federal attention on teacher training. Educators need support to use AI tools without surrendering control of their classrooms. State boards and accreditors must modernize standards so schools can innovate without penalty. Universities should pilot flexible credentialing systems and share results. Industry should make tools and expertise available not only to interns and researchers, but also to K–12 classrooms.

Ethical guardrails must also be built in. Personalized feedback relies on sensitive student data—privacy protections must evolve accordingly. Bias and hallucinations still appear in large language models, so critical thinking must include verifying AI output. And because over-reliance on instant answers can undermine perseverance, educators should sometimes withhold AI to allow for productive struggle.

The window for reform is narrowing. Students already live in an AI-saturated world; education is the laggard, not the pioneer. If schools don’t evolve, we risk raising a generation fluent in prompting chatbots but lacking the deeper reasoning, creativity, and moral discernment that define real learning.

But the opportunity remains. We can build an education system where humans flourish alongside machines—combining AI’s speed with human curiosity, insight, and ethical judgment. Doing so will require strong leadership from teachers experimenting with new pedagogies, from policymakers updating obsolete systems, and from tech companies building tools that enrich rather than erode learning.