AI in education needs more than innovation–it needs intention

Key points:

• Thoughtful design creates AI tools that deepen learning
• How 4 districts use AI tools to transform education
• Easy to find, not always true: Helping students evaluate AI-generated content
• For more news on AI in learning, visit eSN’s Digital Learning hub

Generative artificial intelligence is everywhere, including our schools–and it has the potential to transform education. These platforms offer unprecedented opportunities to personalize learning, refine curriculum development, and support teaching. But as AI adoption accelerates, one critical question remains: Will AI strengthen learning, or will it undermine it?

The future of AI in education must be a shared effort across curriculum providers, district and school leaders, and policymakers. We cannot allow the edtech fervor that peaked during the pandemic to overtake the possibilities afforded by these novel technologies: too much focus on the “tech” at the expense of the “ed.” This AI era is not solely about technological advancement–it’s about ensuring that innovation serves the needs of teachers and learners.

Without careful guardrails, introducing and implementing AI into classrooms risks undermining the very foundation of effective learning. Many AI models are built for general use and later “adapted” for education, but this retrofitting approach can introduce bias, misinformation, and misalignment with instructional goals. If AI is not purpose-built for education, it can amplify instructional inconsistency, widen learning gaps, and erode teacher trust.

AI must align with the science of learning, or it will diminish the technology’s impact on students overall.

AI is changing education–HQIM must be the anchor

Over the past decade, efforts to implement High-Quality Instructional Materials (HQIM) and evidence-based strategies in K-12 schools have begun to deliver measurable improvements in student learning. The “Mississippi Miracle”–a term used to describe the state’s dramatic gains in literacy–and promising 2024 NAEP results in Louisiana, Philadelphia, and Los Angeles illustrate that a commitment to research-backed curriculum can help narrow the achievement gap, though it requires investment and focused effort.

Research from the Johns Hopkins Institute for Education Policy found that access to high-quality instructional materials has a direct impact on student achievement, particularly for historically underserved populations. These curricula support teachers with proven lessons that are created with intention and a clear purpose. Yet, even when states and districts adopt HQIM, teachers continue to supplement the chosen curriculum with external resources–often unvetted, disparate, or misaligned.

Studies show that teachers spend 7-12 hours per week searching for or creating instructional materials, pulling from state websites, teacher-created worksheets, and free or purchased digital content. This mix-and-match approach can result in inconsistent pedagogy, gaps in standards coverage, and lost instructional time. The introduction of AI-powered solutions may also introduce inconsistency or misalignment.

There is growing enthusiasm for using AI to better support teachers–a promising shift that could save educators time and enhance student engagement. According to a 2023 RAND Corporation study, the top five AI use cases teachers identified would help them manage many of the more mundane and time-consuming aspects of their work:

• Supporting students with learning differences
• Generating quizzes and assessments
• Adjusting content to be an appropriate grade level for students
• Generating lesson plans
• Generating assignments (e.g., worksheet materials)

However, most AI tools in education today are not designed with curriculum integrity in mind. The result? Teachers may unknowingly generate content that conflicts with their district’s curriculum, eroding the consistency and effectiveness of HQIM, and unintentionally widening learning gaps rather than closing them.

Without proper safeguards, AI-generated materials could amplify, rather than reduce, instructional inconsistencies–requiring educators to spend even more time reviewing and correcting misaligned content.

To ensure AI strengthens, rather than undermines, effective teaching and learning, education leaders are demanding that implementations are grounded in trusted, research-backed curriculums and aligned with state standards and district priorities. As an example, in its Fall 2024 guidance, the Louisiana Department of Education stated: “Integrating AI technologies should maintain the integrity of high-quality instructional materials used for instruction.”

This underscores a key reality: Education leaders want AI to enhance their chosen curricula, not replace them. Districts and schools invest time and money to identify instructional materials that align with their standards and preferred pedagogical approach, and fit best with what their community, teachers, and students need. They spend valuable time training and supporting teachers to ensure the curriculum will deliver the outcomes they intend. AI-generated lesson plans, practice activities, and instructional recommendations must reinforce the effective implementation of HQIM. By ensuring AI is curriculum-informed, districts can unlock AI’s potential without sacrificing instructional quality.

To benefit K-12 teaching and learning, it is essential that we find ways to propel innovation while minimizing risk. A solid approach to responsible AI ensures generated content is safe, accurate, and academically sound.  Such an approach leverages trusted, research-backed HQIM and employs rigorous vetting for safety, accuracy, ethics, and academic integrity.

This combination helps educators minimize the risks of irrelevant content, misinformation, and bias, and aligns AI outputs with educational goals. And students benefit from a secure, controlled environment where they can safely engage with powerful AI tools and content. This also enables us to safeguard the intellectual property of our curricula and the IP of others, such as trade publishers, authors, and artists, from whom we license content, while maintaining our educational content’s integrity.

AI can help, but only if its use is intentional and aligned with curriculum. Time savings must not come at the expense of instructional integrity, and personalization must keep students on track, not lead them away from core learning objectives. Above all, educators must be empowered when they use AI-driven content, not further overwhelmed.

We are at a pivotal moment. The question is no longer whether AI will influence classrooms. It’s how we ensure it does so in service of teachers, students, and the standards that matter most. Because thoughtful design–not novelty–is what separates tools that deepen learning from those that distract from it.

The future of AI in education won’t be determined by algorithms or innovation alone. It will be shaped by the values we embed in the systems we build and the choices we make today. The opportunity is here. Let’s build AI that not only learns–it understands what learning really takes.

About the Author:

Sari Factor is the Vice Chairman and Chief Strategy Officer at Imagine Learning.