Educators have a complicated relationship with artificial intelligence.
On the one hand, AI-driven tools have made creating lesson plans, summarizing meeting notes or reports, and translating texts quick and efficient. AI can also sort data, predict patterns, and help with the weekly districtwide newsletter. But the technology has sparked some real fears about cheating and misuse by students or even staff.
Data about how much teachers, school leaders, and superintendents use AI in their day-to-day work have been scarce. Two new reports—a study by the research group RAND Corp. and a nationally representative survey of educators from the EdWeek Research Center, conducted in December—attempted to answer this question.
The RAND report, which draws from surveys of more than 9,000 teachers and 3,600 principals, found that almost a quarter of teacher respondents use AI for classroom tasks like leveling texts, while 60 percent of principals who responded said they use AI for administrative tasks like writing emails and newsletters, or drafting or amending school policies.
Similarly, a common thread across qualitative responses from principals in the EdWeek Research Center survey was their use of AI to draft communication. (See how teachers who responded to the EdWeek Research Center survey said they use AI.)
Principals say they use AI to generate or edit emails, specifically for help with grammar and wording. Some principals indicated that they use AI tools to get “content-specific” ideas to share with their teachers; others are using it to collect feedback.
A few principals said they used AI to generate lesson plans for absent teachers, create leveled texts, or develop professional development ideas for their teachers.
One high school principal from North Carolina broke down how they use different AI tools in their work: “I use NotebookLM to create prompts and models based on an uploaded text. I [have] used it to create a study guide to help beginning teachers understand how to deliver lectures with critical thinking. I use Google’s Learn About to find primary documents. I use Gemini to create model paragraphs and create specific syntax.”
Responses and data such as these give states and districts a clearer idea about how teachers and principals are using AI, and what training or guidance they might need, said Julia Kaufman, a senior policy researcher at RAND, and co-director of the American Educator Panels that track data on teachers and principals.
Education Week spoke with Kaufman about RAND’s findings, and what they mean for teachers and principals who want to use AI to better student learning. This interview has been edited for length and clarity.
What are the differences in how teachers and principals use AI?
When [teachers] talked about use of AI, instructional planning was the biggest use. They [had] a huge range of queries to get teaching resources, generating lesson plans, teacher resources, various worksheets and assessments, and even support providing students with feedback and grading.
A very common use was generating text at different reading level than what was they were originally written at, so that students would be able to access texts that were at a simpler reading level.
About one-half of the principals said they were using it to draft communications to stakeholders. They talked about using it to complete other kinds of administrative tasks, like writing recommendation letters, school policies, school handbooks, plan intervention programs.
Fewer principals reported using AI to support teachers. About 15 percent of the responding principals described using AI to support either teacher hiring, evaluation, or professional learning. A small group of principals talked about using AI to help them craft activities for professional learning, like resources for coaches, or slide decks, et cetera. But the more common use was for drafting communications.
Did principals and teachers indicate they want better guidance and training in using AI?
We didn’t ask teachers about the training that they got for using AI in this particular cohort, but we did ask principals [about] concerns that influenced their use of AI.
Between a half and three-quarters of principals said that the lack of professional development plays into whether they use AI, as well as a lack of guidance on the use of AI, and concerns about their ability to use AI tools competently. A majority of principals said that all those things play some role in determining their use of AI.
We did ask principals whether their school or district provided guidance on use of AI at all. It was low. Eighteen percent of principals said their school or district had provided any guidance on usage.
This [guidance] was lower in higher-poverty schools. Thirteen percent of principals in the highest-poverty schools said they got guidance, versus about a quarter of principals in lower-poverty schools.
Is this because high-poverty schools have other priorities for professional development?
I think that’s a safe assumption. Teachers in higher-poverty schools are less likely to use AI, [as are] principals in the highest-poverty schools. So that probably drives whether the school system or the school itself is like, “Oh, we really need to provide guidance.”
One other interesting small point is that although teachers in high-poverty schools were less likely to use AI, when they did use it, they used it more frequently than other teachers. That’s a twist, and suggests that they’re finding more uses for it, so maybe they should be getting a lot more guidance and support to use it.
[A similar analysis on how frequently principals in high- and low-poverty schools use AI was not done for this report.]
What else are you curious about when it comes to AI usage among educators?
I think these data are suggesting that AI is becoming a real use for people, but I think the next step is to know in what ways is it useful. Is it useful for improving efficiency or for improving teaching and learning? When you Google AI tools for lesson-planning, a huge number of things pop up. Are these providing teachers with high-quality lessons? Likely not. They’re providing them with lots of variation in what teachers do with the classroom, rather than help hone in on high-quality materials. I think studying these things more is imperative.
I think schools and states have an extremely challenging job to guide AI, but with very little evidence. What’s important right now is to support research to understand which of these uses are effective, and for what.
There’s nothing wrong with improving the efficiency of what principals and teachers do. That’s great. That’s a good goal. The main thing we push in this report is that any uses of AI, any development of AI, any research on AI, should really be leaning into whether it improves teaching and learning. And I just don’t think we know nearly enough about that to even support all the different uses that are being described, especially among teachers that are using it for instructional planning.
The question for principals is, is AI helping them do the administrative stuff so they can lean into the other stuff that’s really important? Ideally it would help them lean into tasks that are too complex for AI, like supporting teachers in classrooms.
