Empowering K-12 Districts: Navigating AI Adoption for Enhanced Educational Efficiency and Effectiveness

Low AI Adoption in K-12 Education: Lack of Training, or Something Else?

Despite media praise, AI’s integration into schools remains limited, reflecting the slower pace of technology adoption often seen in K-12 settings compared to industry standards. While several surveys have explored teachers’ use, knowledge, and perceptions of AI, there remains a significant gap in understanding administrators’ thoughts, knowledge, and actions—or intentions thereof—regarding its integration to support their work.

According to a Houghton Mifflin Harcourt survey of 1,000 teachers and 215 administrators, just 10% of K-12 teachers utilized generative AI during the 2022-23 academic year, with an additional 38% reporting that they intended to adopt it in the following school year (Lucariello, 2023). However, a Rand research report, released at the end of the 2024 school year, concluded that only 18% of the K-12 teachers in their sample reported using AI for teaching in the 2023-24 school year and another 15% indicated that they had tried AI at least once.

Given AI’s potential to enhance district operations, from recruiting and hiring new staff to teaching and learning, why are K-12 workers hesitant to embrace and integrate these new technologies? One possibility for the low and slow-growing adoption rate is a lack of knowledge and training. Though little has been reported about K-12 administrator knowledge, the same survey from Houghton Mifflin Harcourt revealed that only 1 in 5 teachers felt adequately prepared to use generative AI, such as ChatGPT, in their classrooms and that most teachers (58%) expressed a desire for professional development and/or coaching in AI usage. Another survey highlighted that 87% of teachers had not received any formal training on AI integration, and only 17% felt confident enough to use AI in their work or to educate students on its usage (Prothero, 2023).

Further, though administrators have expressed more hope than teachers about the impact generative AI could have on teaching and learning, many educators do not yet view AI as a useful tool. Only 12% of teachers considered AI tools “very” or “extremely” helpful. Among the limited applications seen as beneficial, generating worksheets (51%), lesson plans (48%), and writing prompts (41%) topped the list (Lucariello, 2023).

Another significant hurdle to AI adoption in K-12 environments is the blocking of platforms like ChatGPT by some districts, citing concerns over student data security (Merod, 2023). Laws in many states require districts to write contracts for student data security to be signed by third party service providers that collect and store student data outside of an internal network. This includes personally identifiable information that is not public, like email accounts, names, and search history. Only service providers that agree to the terms are to be approved and adopted by districts. Large language models, like ChatGPT collect student data but have not been cleared for use, putting school districts at increased legal risk.

It's generally held that educators, including administrators and their staff, are on board with AI and that low adoption is attributed to lack of training and conditional barriers outside of their control, like data privacy concerns and legal risk. However, one missing piece of data that could help clarify the adoption lag is the attitudes of educators towards AI. How do K-12 leaders and teachers feel about AI in the school setting?

K-12 Lens: Providing a Clearer Picture of Educators’ Attitudes Toward AI

Teacher AI Knowledge and Stances

In August 2023, the Institute analyzed data from a survey administered to middle and high school teachers in New England public schools. Participants were asked to select a number on a Likert scale indicating their overall comfort level with AI, such as ChatGPT. A “1” indicated “least comfortable” and a “5” indicated “most comfortable.” The results are plotted on the number line, depicted in Figure 1, below.

Figure 1. Teacher’s Self-Reported Comfort Level with AI, Such as ChatGPT

The most frequently selected answer option was “1” signifying that, during that period, teachers generally felt extremely uncomfortable with AI technology. Among the sample, math and science teachers emerged as the most comfortable group, reporting an average comfort level of 2.1. Conversely, special educators displayed the lowest comfort level, averaging 1.4. Across all content areas, the average comfort level stood at 1.6.

Analysis of open responses revealed that teachers’ knowledge of AI primarily stemmed from second-hand experiences rather than personal experimentation with applications like ChatGPT. Their responses indicated an emerging understanding, largely tied to their immediate circumstances:

Teacher AI Knowledge

• Students are already accessing and using AI
• It can help with traditional assignments, like essay-writing
• It will challenge them to rethink traditional learning activities and assessment measures

Response patterns indicated that teachers wanted to learn more about the following:

1. How to detect AI-generated assignments
2. How to apply AI to the classroom productively
3. How AI will impact teaching and learning in the different content areas

After coding the open responses, three district camps of stances toward K-12 AI adoption emerged: Reluctant Acceptors, Principled Opposers, and Uninformed and Undecided. Each stance is defined in Table 1 and illustrated by quotes from the data.

Table 1. Teacher Stances Toward AI Integration

A follow-up question asked teachers: “Should districts block AI, like ChatGPT?” Teachers were asked to select one of three options that each mapped onto a stance. The results are displayed in Table 2. They indicate that most teachers in the sample fell into the Uninformed and Undecided category. About one-third of teachers were classified as Principled Opposers and one-fifth as Reluctant Acceptors. Not a single teacher expressed enthusiastic optimism for the new technologies.

Table 2. Percentage of Teachers by Stance Toward AI

AI Knowledge and Perceptions: K-12 Leaders

To address the gap in data on administrators’ AI knowledge and attitudes, as well as explore other themes including avenues for improving district operations, the Institute administered its inaugural K-12 Lens Survey. Nearly 700 school leaders selected the option that best matched their stance on integrating AI into the K-12 setting. In comparison with the teachers’ results from the pilot study, the district leaders had a much more positive view of AI. Compared with only 21% of teachers, a slight majority (41%) of K-12 leaders indicated some level of support, with 38% expressing neutrality and 21% expressing opposition. See Figure 2 for the complete breakdown of responses by stance.

Figure 2. K-12 Leader Attitudes Towards AI Integration

Certain factors, like job role and district size appear to have influenced the attitudes of respondents. For example, among K-12 curriculum and instruction district leaders, support for AI stood notably higher than average, at 53%. However, district leaders in small districts, those with fewer than 250 students, reported below-average support, at 29%. Opposition levels hovered around 20% across different job roles and district sizes. See Figure 3 for percentages of respondents by stance, role, and district size.

Figure 3. Percentage of K-12 Leaders by Stance, Role, and District Size

Analysis of open-ended responses reveals that although K-12 leaders may have self-selected into a particular stance bucket, their actual attitudes are more nuanced, even more so than the teachers’ attitudes were just a few months prior. A thematic analysis of more than 300 responses produced a wide range of AI attitude profiles within the three broad categories. See Table 3 for descriptions of each and quotations from respondents to illustrate each category.

Table 3. Educator AI Adoption Attitudes

Four different profiles of opposition towards AI K-12 adoption emerged from the data.

Opposed

Neutral

Uninformed/Undecided

Like the teacher respondents, most district leaders’ neutrality stemmed from their perceived lack of AI knowledge, training, and experience, in general and within the K-12 setting. Many indicated that this has prevented them from forming an opinion on the integration of AI into the K-12 setting. See the responses below:

• “I’m not sure about AI”
• “I need more training in this area to make an assessment, but I suspect that we will have to learn to adapt and find ways for AI to be useful.”
• “I don’t know enough at this time to officially have an opinion.”
• “My concern is that integrating AI will lead to a privacy invasion that the companies aren’t being transparent about.”

Support

Though the largest portion of respondents indicated that they support AI adoption in the K-12 setting, the open responses suggest that “support” might mean different things to different district leaders. See Table 4 for the variety of stances that emerged from the data.

Table 4. Range of K-12 Leader AI Support Stances

AI Supporters’ Plans for Integration

District leaders supporting the integration of AI into K-12 settings outlined various use cases in their responses to the K-12 Lens. These responses spanned domains including teaching, assessment, administrative work, and student learning. See Figure 4 for examples from the data. Note that these are not all in-use but rather ideas for how AI could provide benefits to those in the K-12 setting.

Figure 4. Potentially Beneficial Use-Cases from K-12 Lens Respondents