Two short years ago, the technology-in-education problem boiled down to one culprit – the  smartphone. Consensus quickly emerged then that smartphones were distracting students and deterring them from learning. (Research on how much of a distraction they are is a decidedly mixed bag.) 

Virginia became among the first states to require a ban on the devices during instructional time. Problem identified. Policy installed. Not a perfect solution, but a reasonable one.

The struggles over what to do with artificial intelligence (AI) in the classroom have no parallel with the smartphone challenge.  

There is little consensus about AI’s proper role in learning. And unlike smartphones, simply banning it isn’t an option – students still will turn to it after school for their education.

As classes gear up across Virginia this week, districts’ approaches to handling artificial intelligence range from cautious and limited to aggressive moves that fully integrate the technology across the division. 

While there is no one course of action, the commonwealth is coming to some agreement about the challenges. And that is driving innovation. 

Facing the Oncoming Storm

In early June, when teachers are typically settling into their summer sleeping habits, Stafford High School played host to a “Region III AI Leadership Summit.” 

Among the 400 teachers showing up early for the event was one who said aloud what many likely felt – “I’m scared of this technology,” she said over breakfast in the commons area. 

It’s a fear that Sharon Shewbridge, chief learning and innovation officer for Stafford County Public Schools, understands well. 

Teachers “like to be the experts in the classroom,” she said in an interview with the Virginia Free Press (VFP), “and if we don’t know what the tool is or how [it] can be used, we’re a little skeptical and fearful when the kids know more than we do.” 

The summit was, in part, a move to remove some of that fear. 

Among those most concerned? English teachers, said Shrewbridge. Teaching writing and literature is an exercise in balancing evidence and conflicting ideas into a tightly argued essay that reflects a clear line of thought. AI improperly used too early in learning to write can deprive students of building those skills. (See sidebar story: “Learning with Artificial Intelligence.”)

Simply teaching students how to use AI as a tool won’t resolve that problem.

The “missing piece,” Shrewbridge continued, is “How are we teaching students ethical and effective use of artificial intelligence?”

“Students and staff have to have foundational knowledge and skills before they can use AI,” Stafford Schools Superintendent Daniel Smith told VFP. That has been a focus of the division coming into the new school year.

That belief led to Stafford County School over the summer instructing teachers in the district’s “frameworks for AI and how we think about it here in Stafford,” Smith said. 

This fall, when high school English teachers enter their classrooms, they’ll take that knowledge with them as students work through eight learning modules that will educate them in the ethical use of AI. The modules were put together by University of Mary Washington (UMW) professor Anand Rao, who serves as director of the university’s Center for AI and the Liberal Arts. 

Rao, who developed the Intro to AI course at UMW that has been taught to more than 500 students to this point, was approached by Smith about adopting the course for high school students.

“[Smith] and Dr. Shebridge had the vision to include AI literacy within their AI plans” for Stafford County Public Schools, Rao told VFP. 

Rao continued, stating that “Smith has the sequence right.” There’s a version of AI, Rao said, “that spares students the effort of thinking, and a version that raises the ceiling on what they can think about.”

The only thing, Rao said, that separates them is “whether the student learned to do the work first.”

Do We Value This or Not?

While it’s easy to remove the threat that smartphones present educators – distraction in the classroom – AI cannot simply be removed. Beyond the walls of the classroom, there are no boundaries on how often or how much students may turn to it.

This creates a number of challenges for educators, not the least of which is the issue of student cheating, and what it means for learning. It’s a challenge that is pushing the discussion from tools and guidelines to ethics and the value of learning. 

Adam Garry, who was the keynote at the June conference, talked with the Virginia Free Press (VFP) about a panel at the AI Summit that included students. Cheating was a key talking point during that panel. 

“My takeaway” from that session, said Garry, who formerly led Dell Technologies’ education and innovation strategy and today is president of StrategicEDU Consulting, is that “we need to define cheating. There were things the kids [on the panel] were saying that I bet people in the audience thought were cheating,” and the kids on the panel did not.

The solution, he said, is to “frame [talking to students about AI] around academic integrity.” 

That point hit him as one student on that panel, according to Garry’s summation, said: “When it comes to cheating, we make a decision. Do we value this [work] or not?” 

Garry points out that following school, after-school activities, and possibly work, students face a couple hours of homework at 10 p.m. These students, he said, make a decision about what matters as a learning point, and what is “busy work and [they] know it’s a game to get a grade.” 

“If we don’t talk to them about values, and we don’t really help them understand why learning is super important,” he continued, then students are more likely to quickly turn to AI rather than “having that internal dialogue” about what matters. 

While estimates vary, there is a consensus that well over half of students in K-12 are using AI for graded work.

That worries Cody Christensen, an instructional technology and resource teacher in Colonial Beach Public Schools. “There’s always a subset of kids (about 15%) who are going to cheat,” he told VFP. The problem is that AI has “made cheating so much easier. You’re not just copying from Wikipedia, but you’re getting a new answer. It’s a bit too tempting for kids.”

Wrestling with the Alligator

Christensen understands the need to talk with students about the importance of learning. 

Today’s students will need to use AI in life, he told VFP, “but there is also this thing called “creating imperfect things” and learning to build on these these going forward.”

He has a phrase for this – “Wrestling with the alligator.”

“If I just ask ChatGPT to do things for me,” he said, “I’m not going to learn to research, learn to learn, or [learn to] look for things.”

Administrators, by way of contrast, have their own alligators to wrestle with. Among them, getting teachers to understand that AI is here to stay. “We’re not going to strongarm teachers into using it,” Christensen said, “but the tool is not going away. And the kids are figuring out AI long before the teachers are. The kids are on the frontlines. They know how to use it. If the teacher doesn’t,” they’re going to face a significant uphill struggle.

One way to get teachers on board, he believes, is to help them apply AI to their work days. “Teachers spend as much time out of school [working] as they do teaching,” he said. “Let AI do 80 percent” of the work that it can realistically handle, and teachers will begin to see the benefit.

“If we start from a place whereF the teacher is smarter than the AI,” Christensen said, “then we are in a better place.”

Unequal Distribution

In some ways, Virginia is well ahead of curve in helping districts learn to work with AI and integrate it into operations as well as the classroom, according to Garry.

A new state law that went into effect on July 1 requires the Virginia Department of Education (VDOE) to create statewide guidance on using AI in K-12 settings. Further, school boards are required to adopt policies in line with that guidance. 

But the state was moving ahead on AI two years before. 

In August 2024, for example, the Virginia Association of School Superintendents partnered with VDOE to bring together some 400 education professionals across 66 districts to deliver AI training.

Garry told VFP that the idea for the meeting came after Dave Myers, then Chief Information Officer and Assistant Superintendent for the Department of Data, Research, and Technology with VDOE, approached him and said: “Talk to me about what we can do at a state level to really support our districts.” 

The program that emerged showed these educators how to “build all the things they needed to roll things out in a safe and secure way,” said Garry. 

“Based on that work,” he continued, “they were prepared to be innovative.”

Despite this, however, AI rollout remains uneven across the commonwealth.

Garry sees two problems. One is that while the state helps pay for districts to provide instructional technology resource teachers, like Christensen, a significant number are focused on tools, which doesn’t necessarily help with bringing teachers along. Tools training, Garry said, “isn’t helpful.”

The other is that some districts are missing an opportunity by focusing their attention on AI to the teaching and learning environment, and not the district’s overall operations. 

Money, he said, is not the issue. “If [districts] want to use foundational models like Gemini and CoPilot,” he told VFP, they are already paying for that. They don’t need to go out of pocket for other tools.” Schools can get what they need, he concluded from Google Classroom or CoPilot, and from teaching people how to use the frontier models. 

Consequently, how successful AI is being rolled out depends on the district.

A recent report from Virginia Commonwealth University’s Metropolitan Educational Research Consortium (MERC) highlights the inconsistencies.

“School systems are still trying to wrap their heads around” AI use, said David Naff, one of the report’s lead authors and an associate professor of research, assessment and evaluation in the School of Education, as well as the director of MERC. “There’s a lot of variability right now, and there could be some benefit from more consistent and central guidance.” 

Among the report’s findings is that “Staff Use and Training and Literacy are the domains with the least existing infrastructure and require priority attention, including investment in AI-specific professional learning.”

Human Evolution vs. AI Evolution

Few educators would dispute that AI is here to stay. But learning to deal with it in an educational setting is forcing some uncomfortable questions that go far beyond the tool itself. 

To begin, schools are being forced to think about what really matters in terms of the material being presented to students. It’s a discussion that is long overdue, following a quarter-century of outcomes based learning that leans heavily on objective, multiple choice exams. 

Many students do know the difference between learning that matters and learning that is simply busy work whose only benefit is a grade. 

Districts that are tackling this by leading with the ethics of AI use, and talking honestly with students about “wrestling with the alligator,” are ahead of the curve. Those still looking at AI as a tool are missing the fundamental shift that AI is bringing to student learning. 

Christensen, like Garry, believes Virginia has been doing a pretty good job to this point. The challenge? “We are moving at a very human pace,” Christensen said, “but [AI’s] moving faster than we are.”