The number of districts that trained teachers on generative artificial intelligence more than doubled from 2023 to 2024, according to a new analysis by the research and policy firm RAND Corp.
In the fall of 2024, 48 percent of districts reported they had trained teachers on AI use—up from 23 percent in the fall of 2023. RAND surveyed about 300 districts in 2024 and about 200 in 2023 and weighted their responses to make them representative of all public school districts.
If districts follow their plans for the current school year, nearly three-quarters will have provided AI training to educators by the fall of 2025.
This professional development has been slow to gain ground in schools. While generative AI became mainstream in late 2022 with the launch of ChatGPT, 7 in 10 teachers had not received any training on how to use AI in the classroom by spring 2024, according to a nationally representative survey from the EdWeek Research Center. By fall 2024, though, 43 percent of teachers said they had received at least one training session on AI, the EdWeek Research Center found.
Both teachers and principals are increasingly using AI in their day-to-day work. While principals use AI tools for administrative tasks like making schedules, writing newsletters, or crunching large datasets, teachers are more likely to use AI to generate lesson plans, customize lessons for students who might learn at different speeds and levels, or even use it creatively—to emulate a historical or fictional character to teach a particular lesson, for instance.
Even so, most districts took a gradual approach to introducing their teachers to AI.
In addition to fielding the surveys, RAND conducted in-depth interviews with 14 district leaders on how they crafted their AI trainings in 2024. Instead of jumping straight into AI tools and their use, district leaders focused first on tackling teachers’ concerns about the technology.
Of the 14 leaders interviewed, all but one said they encountered “negative views of AI and reported that some teachers view it as a threat to traditional teaching methods or a tool for student cheating,” according to the report.
District leaders said they provided training on the fundamental workings of AI—and its limitations—to lower teachers’ anxiety, and “shift away from an antagonistic, cheating-centered mentality” about AI.
Among the 14 districts, trainings varied in length and content—from daylong trainings to “bite-sized” content delivered through regular emails and newsletters. All but one district made such training voluntary.
“I am not trying to push AI with my people, but I am trying to give them the tools if they’re interested. I know that as [AI adoption] starts happening, it will grow organically and very quickly,” one district leader told RAND researchers.
District leaders tend to craft their own training
The district leaders interviewed by RAND preferred crafting their own trainings, rather than relying on outside experts, said Melissa Diliberti, an associate policy researcher at RAND and a lead author on the report, in an email.
“This do-it-yourself approach, though potentially employed to meet specific district needs, also reflects a scarcity of external experts who are capable of providing appropriate training,” Diliberti said.
Of the 14 district leaders interviewed, 11 took a DIY approach to crafting their own AI training. They pieced together different kinds of resources put out by tech companies like Google, as well as education organizations like Digital Promise and the International Society for Technology in Education.
Half of the interviewees said they struggled to find training experts who were knowledgeable about using AI in an educational context.
A time and resource crunch also pushed district leaders to pull together their own training, in the absence of ready-to-use resources, the report stated. Professional development in AI jostles for space among several other competing PD priorities, which can affect the length and depth of such training, the district leaders said.
District leaders also used their own “negative” experiences with AI trainings—where they were left overwhelmed—to streamline the training they offered so it addressed teachers’ immediate needs.
A suburban West Coast district, for instance, first introduced “play” sessions to alleviate teachers’ anxiety about AI, before gradually moving onto more instruction-based learning.
The “play” approach, in this district and others, consisted of teachers experimenting with an AI chatbot or platform and getting the hang of what AI can do. There were more focused sessions, too, in which district leaders introduced tools like MagicSchool or ChatGPT to show teachers how to use them to plan lessons or generate differentiated content for students.
Chris Chism, the superintendent of the Pearl school district in Mississippi who was not associated with the RAND report, also created his own AI training module for teachers. His training, too, emphasizes both experimenting with the technology, and showcasing more streamlined, direct uses for the classroom.
In an interview with Education Week last July, Chism said he had created mini chatbots that deliver essential information to teachers about specific topics. One chatbot, for instance, can generate lesson plans aligned with the state’s standards.
While district leaders’ trainings may be more in tune with a teacher’s needs, Diliberti said the quality and what’s covered in these trainings could vary between districts.
There’s an equity gap in AI training
Despite the overall surge in AI training, the RAND report shows that such training is unevenly distributed—compared to nearly 67 percent of low-poverty districts that have introduced AI training for teachers, only 39 percent of high-poverty districts were able to do the same.
RAND used districts’ plans to project what the training gap might look like in the fall of 2025. While nearly all low-poverty district would have trained teachers on AI use, only 6 in 10 high-poverty districts would have done so, if districts stick to their plans.
Schools in low-poverty districts are more likely to have the time and resources to experiment with AI tools and figure out how to use them well for instructional practice. It’s not a luxury that schools in high-poverty districts may have, said Diliberti.
This gap in training, or guidance, on how to use AI persists among the principal group, too, according to a previous RAND report published in February. Thirteen percent of principals in the highest-poverty schools said they got guidance, versus about a quarter of principals in lower-poverty schools.
“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,’” Julia Kaufman, a senior policy researcher at RAND and lead author of the February report, told Education Week at the time.
The faster takeup of AI by teachers in low-poverty districts means that teachers and students in high-poverty ones won’t have the same opportunity to learn from AI tools, the report says.
There are still no clear links between greater reliance on AI tools and higher-quality teaching or learning, Diliberti said.
“But to the degree that we anticipate this might be true, greater uptake of AI in historically advantaged school settings does raise concerns that the rollout of AI into schools could exacerbate longstanding fault lines of educational inequality,” Diliberti added.
The report calls for more targeted funding by federal and state education agencies to high-poverty districts to procure model professional development in AI and cultivate connections to networks of experts.
The report also suggested that districts with more experience in training teachers on AI should share their expertise with other districts through state and regional education networks.
The surge in training notwithstanding, Diliberti said it’s still too early to figure how exactly AI might fit into a district’s PD schedule for the year.
“Evidence from our district leaders suggest that their trainings are in the very early stages. [They are] more focused on just getting teachers comfortable with AI tools,” Diliberti said. “Regular teacher training on whether and how to best incorporate specific AI tools into instructional practice may still be some time away.”
