Evidence Matters | AI in K-12 Education: So What Now?

In Part 1 and Part 2 of this series, we explored AI’s promise as a learning partner as well as the evidence gaps and risks surrounding educational technology. But the question remains: what should we actually do?

The answer is unlikely to be either banning AI or allowing students to use it for everything. Instead, schools need to keep one principle at the center: technology should support learning, not replace it.

The Goal Isn’t Less Technology. It’s Better Technology Use.

Research increasingly suggests that the educational impact of technology depends on how it is used. An OECD review found that simply providing students with digital tools does not guarantee better educational outcomes. Instead, effective use requires appropriate instructional and pedagogical approaches. That distinction matters with AI.

Using AI to explain a difficult concept, generate additional practice questions, or provide feedback can create opportunities for learning. A 2025 meta-analysis of 68 experimental studies found an overall positive effect of generative AI on learning outcomes, while also finding significant variation depending on the educational context and how the technology was used. But asking AI to write an essay, solve a problem, or produce an answer can bypass the very thinking the assignment was designed to develop.

Students Need AI Literacy—and Subject Knowledge

Students will need to understand how AI works, but AI literacy should mean more than knowing how to write an effective prompt. Students also need to recognize that AI can produce confident but incorrect information. They need to evaluate sources, check claims, identify missing context and decide when human judgment is necessary. That makes foundational knowledge more important, not less.

A student who understands mathematics is better positioned to recognize an AI-generated mathematical error. A student who understands historical evidence can question an unsupported claim. A student with strong writing skills can evaluate whether an AI-generated argument is actually persuasive.

As the KnowledgeWorks’ 2026 analysis of the “expertise paradox” argues, AI is increasing the importance of human judgment, oversight, and deep expertise even as automation reduces some of the routine tasks through which people traditionally developed those skills. And this is where the value of teachers may become most visible in the age of AI.

Teachers Still Matter

AI can generate explanations, practice activities, and feedback. But teachers provide something technology cannot easily replicate: knowledge of individual students, classroom relationships, context, and professional judgment. In 2024, the OECD highlighted that while AI can provide some personalization, it lacks the intuition and human connection of teachers. Teachers can recognize when a student is struggling academically or emotionally, provide encouragement, and adapt their support to individual circumstances.

The report also emphasizes the teacher’s role in creating a classroom environment where students feel comfortable asking questions and taking intellectual risks. Thus, the future may not be teachers versus technology. It may be teachers using technology intentionally, while deciding when students should work independently, when AI can assist them, and when using AI would undermine the learning objective.

Learning First

Not only does this approach require schools to rethink assignments—it also requires them to rethink assessments. If AI can produce a polished essay instantly, the finished product may tell educators less about what a student actually understands. More emphasis may therefore need to be placed on producing outlines and drafts, explaining reasoning, defending conclusions, revising work, participating in discussions, and demonstrating skills independently.

Ultimately, the goal shouldn’t be AI-free education or AI-first education. It should be learning-first education. AI may sometimes be able to produce a better answer than a student. But education isn’t simply about producing answers. It is about developing the person who can understand a problem, evaluate information, ask good questions, make informed decisions, and eventually solve problems independently.

The smartest presence in the classroom may sometimes be the AI, but the purpose of education has never been to create the smartest machine in the room. It has been to help the human beings in the room become more curious, more capable, and more independent.

The technology will keep changing. The mission shouldn’t.

---

Kelly Gregory is the Riley Institute’s Director for Public Education Partnerships and Projects and previously taught for 11 years in South Carolina public schools. She holds a bachelor’s degree in Psychology and a master’s degree in Special Education. She also holds a National Board certification as an Exceptional Needs Specialist. She can be reached at [email protected].