Exploring the Current State of AI in Education

Introduction: AI Enters the Classroom

Arti?cial intelligence (AI) is seemingly affecting all industries that we interact with today with promises of widespread disruption, and the education sector is no different. With an eye toward disaggregating fact from ?ction, A/B Consulting conducted in-depth conversations with and administered a survey to decision-makers* and educators to understand both the current state and future outlook of arti?cial intelligence in education.

We settled on seven themes that are likely to change and evolve as administrators and educators begin to understand this emerging technology at a deeper level in the coming years and the impact that it will have on both teacher and student outcomes. Without further ado, the seven themes are as follows:

* Includes Superintendents, Assistant Superintendents, CIOs, CTOs, and CAOs

1. The use of AI in education is in its infancy

Most educational decision-makers are taking a cautious ‘wait-and-see’ approach to understand how AI and AI-powered tools will disrupt their districts. In fact, a recently administered survey by A/B Consulting revealed that while 40% of decision-makers currently use AI-powered tools in classroom and back-of?ce settings, current use cases are largely experimental, occur at the classroom level, and leverage free tools (e.g., ChatGPT). Consequently, a majority (80%) of decision-makers are only allocating a small portion (0-9%) of their annual budgets to AI-powered tools today.

A Software Development Director for a mid-size public charter school encapsulated the sentiment of most districts we engaged with by stating, “First of all, I’ll say that there is nothing that has been sanctioned and rolled out network-wide. Use is more sporadic here and there for individual needs on a more experimental basis.” An Assistant Superintendent for Innovation for a public school district shared a more cautious tone by stating that, “We want to be quick but not in a hurry [to adopt AI]. To that end, we are erring on the side of teacher and staff training and empowerment instead of recommending one tool over the other.”

2. Despite being in being in the early stages, AI adoption in education is viewed amongst decision-makers as a matter of ‘not if but when’

A majority of surveyed decision-makers view their district’s attitude as ‘accepting’* toward AI usage and plan to increase** their budget allocation to AI-powered tools in the next three years (76% and 64% of decision-makers, respectively). Moreover, decision-makers view it as their responsibility to leverage AI to both (1) reduce the burden placed on teachers and (2) proactively prepare students to use AI safely as it becomes increasingly entrenched in their daily lives.

As decision-makers look forward to increased AI adoption, they will be weighing many considerations when deciding whether to implement AI-powered tools. Survey results indicate that student safety, the effectiveness of AI-powered tools in improving outcomes, and the ability to educate staff on how to properly use tools are most important (84%, 80%, and 76% of respondents, respectively).

A Senior Director of Technology & Data Support for a mid-sized charter school summarized a widely shared future of AI in education by commenting that “AI is here whether we want it to be or not. So what can we do to use it for good? I’ve been starting to think about that and talking about it with my colleagues.” A Former Assistant Superintendent for a large public school district shared a similar sentiment by saying “I’ve never doubted that AI would become a transforming piece of public education. It is just a matter of time. I do think that in a number of years we are going to be having a very different conversation about how AI has transformed the educational landscape.”

3. Educators currently using AI-powered tools are in the ‘test and learn’ stage and tinkering with AI for a wide variety of use cases

Widespread experimentation of AI in education is currently underway despite low district-level budgetary allocations to AI-powered tools. Educators are leveraging tools such as ChatGPT, Eduaide.ai, gradescope, and MATHia to enhance everything from lesson plans and grading for reducing administrative burdens to developing personalized learning for improving student outcomes.

Moreover, detailed survey results reveal that there is a high degree of variance regarding how educators are using AI, which aligns to their current attitude of experimentation toward the technology. For example, of 13 use cases surveyed, the most popular AI use case, adaptive learning platforms, is mentioned by only 56% of educators, followed by just 52% for create assessment questions and 51% for generate in-class content.

While most districts we engaged with are early on in their AI journeys, our research revealed a few instances in which AI is being leveraged at the district level in a more mature way. One example is a district in Georgia spanning ~23 schools that is leveraging not only ChatGPT but more custom-tailored AI-powered tools to transform student learning and administrative ef?ciency. In particular, the district has implemented products such as Lexia, Renaissance Learning’s Star, Maya, and Edgenuity to personalize learning paths and enhance literacy, math, and ESL education.

*Percentage of respondents who report their district’s attitude toward AI usage as ‘somewhat accepting’, ‘accepting’, or ‘extremely accepting’

**Percentage of respondents who plan to increase their investment in AI-powered tools ‘signi?cantly’ or ‘somewhat’ in the next school year

4. Despite some early adoption, a majority of decision-makers have expressed concerns with AI-powered tools in education

Three quarters (76%) of decision-makers reported having concerns with AI-powered educational tools. Top concerns include maintaining teacher engagement (i.e., avoiding AI dependency), student plagiarism, data privacy and security, and improper training on AI’s use (84%, 63%, 63%, and 58% of respondents, respectively).

Regarding teacher engagement, decision-makers fear that an overreliance on AI-powered tools may lead to a disconnect between teachers and the learning process and affect education quality. For instance, an Assistant Superintendent of a mid-sized school district described an experience in which a supposed supplementary tool to enhance student learning ended up becoming the de facto curriculum. Teachers have also become overly dependent on automated grading, resulting in a lack of hands-on engagement and a diminished ability to address student-speci?c mistakes.

Regarding plagiarism, decision-makers fear that teachers and students will become lazy and negligent in questioning or checking material generated from AI tools and undermine the learning experience. For instance, a Director of Software Development at a charter school network mentioned “When ChatGPT came out, my whole district wanted to ban it. It was seen as too risky from a plagiarism and academic integrity standpoint.” Similarly, a Senior Director of Technology & Data Support at a mid-sized charter school network said, “If I have an advanced chemistry formula that I need to balance and put it in ChatGPT, I didn't actually learn the underlying components”.

Regarding data privacy and security, decision-makers have become cautious that data being used to inform AI tools will be misused or improperly stored. For example, a Director of Software Development of a mid-sized charter school expressed concern about personal student data: “Am I going to have kids inputting personally identi?able information that is going to be stored someplace and then sold?” Similarly, the Chief Business Of?cer of a small district is concerned about the risk of a data breach involving sensitive information and is taking steps to prevent it: “We would not willingly volunteer data to be in an AI training set. Anything that we thought would reveal con?dential information about students in any way… would be my biggest concern area.”

Regarding improper training, decision-makers fear that a lack of investment in comprehensive AI training programs may hinder the effective use of AI tools in the classroom. An Assistant Superintendent of Innovation at a large school district noted, "We’re training our teachers and staff on how to best use the AI tool themselves versus being lazy and buying a service... That takes skills and takes training. If schools aren’t training like we do, they will be at the mercy of the product vendor."

5. District-level guidance and policy can help mitigate AI concerns but districts are taking a cautious approach

Clear AI guidelines and policy will be critical to mitigating concerns and furthering AI adoption as the education sector continues to experiment with and learn about the capabilities of AI-powered tools. A former Assistant Superintendent of a large school district emphasized that once a formal policy is in place, acceptance, and usage of AI tools will accelerate: “Once the policy is ultimately developed, [usage of] this technology will be far broader than just ChatGPT.”

Discussions are underway at the school level, where general best practice guidelines for AI are being considered, but such discussions often fall short of an approved district-wide formalized policy. A Director of Software Development of a mid-sized charter school stated, “I wouldn't call it policy at this point, but we’ve gotten a group of cross-functional stakeholders together to set some guidance with teachers about what we absolutely should not do with AI.”

One noteworthy piece of AI guidance is the Generative AI Readiness Checklist. This document provides a comprehensive framework for assessing preparedness and ensuring responsible AI integration in educational settings. An educational stakeholder we spoke with referenced the work, stating “I'm intrigued by the work that CoSN (Consortium of School Networking) has done with the Council of Great City Schools.

6. Larger districts are much more likely than small and medium districts to ?rst drive AI tool selection and use at the district

All large districts in our survey report that AI selection will be done at the district level (vs. 60% for medium and 33% for small districts, respectively). Ultimately, large districts have an elevated risk of scaled, negative impact and are unlikely to be the ?rst to test and adopt unproven technologies, opting instead to centralize and control the AI adoption process. In fact, many large districts in New York, Baltimore, and Los Angeles have gone as far as banning ChatGPT to evaluate the bene?ts and risks that accompany the technology.

While large districts will eventually have an active role in driving scaled AI adoption, small and medium-sized districts are likely to drive AI in its experimentation phase at the teacher and school level. A Director of Software Development at a charter school network summarized the role of both teachers and administrators in increasing AI adoption by sharing that “Strategy number one [in past education technology adoption efforts] has been pilots at the ground ?oor and seeing what gets the most excitement amongst teachers, but that does not mean that school systems can’t strategically tend to the garden of growth.”

7. AI-powered tools are more likely to appear as valued add-ons to existing technologies as opposed to stand-alone solutions

Decision-makers view AI as an eventual must-have feature that will enhance the productivity of preexisting solutions. To this end, AI will eventually become table stakes over the next 5-10 years and something incumbents will need to incorporate into their solutions to remain competitive. As a result, ‘AI-powered tools’ as we know them are more likely to appear in a majority of cases as enhanced versions of a district’s current tech stack and embedded into current technology spending as opposed to separate technologies with their own, broken-out budget line items.

A public school district Assistant Superintendent shared an interesting perspective on how AI will be introduced to school stakeholders when he said “We are still exploring AI applications but are ?nding that several educational tools are now incorporating AI into their platforms. I imagine that a majority of our staff will be using some form of AI even if they don't know they are using it.”

Closing Remarks

While only in its early stages, AI in education will bring about disruptions that have the potential to dramatically reduce teacher administrative burdens and improve student outcomes. The current landscape of experimentation is likely to give way to district-wide use of AI to capitalize on a wide variety of scaled, meaningful use cases. The time is now for districts to ensure that strong policies are in place to prepare for the rapid, increased use of AI amongst teachers and administrators and mitigate AI’s many well-founded concerns, from teacher over-reliance of AI to data privacy, data security, and plagiarism.