AI in Education: New Guidance from the Department of Education

What Is the DOE Recommending for AI?

In its report “Artificial Intelligence and the Future of Teaching and Learning,” the DOE digs deep into AI’s potential for improving the classroom experience.

The report “provides some insights as well as recommendations” for the use of AI in education, says a DOE official. “We’re looking at overall thinking about how AI can shift today’s ed tech use in two different ways: from capturing data to detecting patterns in data and from providing access to instructional resources to automating those decisions about teaching and learning processes.”

The report spells out a number of practical strategies, including:

• Keeping humans in the loop: “The highest priority is emphasizing humans in the loop,” the official says. “We want to make sure that educators are involved in all aspects of this integration into our classrooms.” Teachers should be included in procurement discussions as schools take on new AI-supported technology. “We also want to make sure that the humans know which tools are using AI,” she adds. “They should know how student data is being used and that they can then turn that on or off.”
• Developing new models: Rather than automating existing paper-based processes, AI offers the chance to create new pedagogical models. “Are we asking for rote memorization? Or are we really preparing students to think more critically and look at inquiry-based and project-based learning?” the official says. “We can look at the different types of instructional models” with the support of AI.
• Strengthening trust: Teachers and students will need to have a high level of trust in AI products. That means schools must help them understand how these tools work. “Do we know when it’s being used? Do we know where the data is coming from? Do we understand the algorithms being used? We certainly don’t want to put ourselves in any sort of situation where we are using algorithms that might bias students,” the official says.

How Does the DOE Report Compare to Other Guidance on AI?

In addressing potential bias in AI, the DOE report builds on other recent AI-related guidance documents.

The White House’s Blueprint for an AI Bill of Rights addresses ethical ground rules. It calls for AI to be developed “with consultation from diverse communities, stakeholders and domain experts to identify concerns, risks and potential impacts of the system.” This helps to ensure equitable use of AI capabilities.

WATCH NOW: Merlyn Mind releases the first large language model for education.

In looking at equity, the DOE is talking not only about algorithmic bias but also about access to educational technologies.

“Prior to the pandemic, the more affluent schools had more access to connectivity and devices. We don’t want that to happen” with access to AI, the DOE official says.

Other recent guidance, meanwhile, addresses some of the technical aspects of AI implementation. Google’s Secure AI Framework, for example, aims to support the development of secure AI systems. It calls for a strong security foundation in the AI ecosystem, including early detection of AI-related cyber incidents, automated defenses and platform-level controls.

What Can K–12 Leaders Take Away from the DOE’s Guidance?

The mere fact that the DOE has stepped up with guidance is a positive sign, Roschelle says.

“The European Union, the U.K., Australia — all these places have been producing statements, reports, policies. It would be strange if we could read reports from all these other policymaking organizations, and there were none from the United States,” he says.

“This is an important discussion that we need to have. People need those conversations to have a starting point,” he adds.

As teachers and school officials start to have those conversations, they can take away key points from the DOE guidance, the DOE official says.

First, AI has a transformative potential. “It enables new forms of interaction. It’s supporting educators via assistants and agents. It can help address variability in student learning with more powerful forms of adaptivity,” she says.

Second, it’s not going to take jobs away from educators. “We reject the notion of AI replacing teachers,” she says. Rather, it will free them to do their jobs better.

“We know that we certainly can reduce the overall burden on teachers, and then hopefully give them more time to provide personalized and individualized learning,” she said. “An electric bike amplifies the human power to get you somewhere. That’s how we envision AI being used in education.”

Roschelle sees big potential here. “There is a lot of routine work that goes into being an educator. You have to write lesson plans. You have to file individualized student learning plans,” he says. “If we can help educators with some of those things, they could spend less time on that and more time with students.”

This, in turn, should help with recruitment and retention. “It’s hard to retain high-quality people in the profession right now. We could make this a more attractive job,” he says.

For that to happen, districts will need to leverage the DOE guidance and other emerging ideas to formulate new policies — policies that empower the use of AI while still leaving teachers firmly in charge. “We have to be clear that the responsibility for what their students are learning, what experiences they’re having, stays with the educators,” he says.

UP NEXT: How can schools ensure equity for all K–12 students?