The challenges associated with the use of AI by students are not completely “unprecedented.” Many of the discussions and arguments about AI echo the ones about calculators. On one end of the spectrum are those that claim the existence of these technologies means that there is no need for students to master the skills these tools can automate. On the other end are those who believe that students should not be able to use these technologies at all until a certain level—often college. Most teachers find themselves somewhere between these two extremes.
I lived this tension with calculators over many years of teaching math at the middle school level. After trying several different approaches, I settled on a principle. Calculators may be used on any work where the primary learning goal is not computation. In other words, when the learning goal is adding fractions, calculators cannot be used. When the learning goal is finding averages, calculators can be used. While not perfect, it did work well. I had a clear way of defining when students could use calculators and why.
I still had some decisions to make, though. One learning goal is calculating surface area and volume for cylinders. While it is a “computation” goal, students would spend significantly more time on repetitive multiplication than working with the actual formulas. In the time it takes to do one problem by hand, a student might be able to do 3 or 4 with the calculator, allowing them to put more focus on the formulas, especially the proper application of the order of operations.
This made assessment more precise. The large number of calculations in a single problem increased the chance of a careless error. By using calculators, those errors were decreased allowing me to focus on whether students understood the procedure. To get a correct answer, students also needed to understand the formulas to properly enter values into the calculator. We regularly reviewed and practiced computation throughout the year so doing them here by hand was not necessary. This was actually more effective because computation topics were not equally represented in learning goals that were not directly about computation.
My calculator policy was at the forefront of my mind when I was asked to draft an AI policy for my school. Like any other academic or behavioral policy, it needed to be specific enough that violations could be easily defined but also general enough to avoid listing all acceptable and unacceptable uses.
The AI policy was a restatement of my calculator policy. AI should not be used when it is doing the work of a learning goal. This was meant to address the biggest challenge we were facing, students using generative prompts to produce entire written responses to questions in English Language Arts and Social Studies. Consider the following pair of Common Core standards.
CCSS.ELA-LITERACY.W.6.1.A
Introduce claim(s) and organize the reasons and evidence clearly.
CCSS.ELA-LITERACY.W.6.1.B
Support claim(s) with clear reasons and relevant evidence, using credible sources and demonstrating an understanding of the topic or text.
Since the learning goals are to introduce a claim and organize the evidence to support that claim, the results of a generative AI prompt would not be allowable. This is the equivalent of using calculators to practice computation. Other AI tools might be allowed.
Spell and grammar checks are a form of AI that might be allowable in this scenario. They would not aid students in the performance of the primary learning goal. Like the calculator in the cylinder problems, they would help produce a response while still requiring students to work with the learning goal.
Underlying this approach to AI policy are two simple ideas. First, students learn and grow by working through these assignments and assessments. When AI takes away the work and the struggle, it takes away the learning as well. Second, AI can be used as a tool to support learning in ways that do not take away essential work and struggle.
To give teachers room to experiment, the policy also explicitly stated that teachers were free to design assignments that integrated AI in any way they thought had value. While essential for our technology teacher, who was responsible for teaching students how to use these tools appropriately, it also allowed any teacher to experiment with the technology’s benefits within a defined policy.
The policy was not perfect; no policy is, but it kept the focus on the essential role of student effort and struggle in the learning process while acknowledging and taking advantage of the benefits of a new tool. I do not know whether the policy as I drafted it still exists, but I hope the principle does. Tools that help students focus on particular learning goals have benefits and should be used. Tools that take away from student struggle on specific learning goals should be avoided. This is as true for AI as it is for calculators.
The thinking in this article reflects the principles of The Core Teaching Framework, especially the practice of planning and instruction as laid out in Navigating Instructional Flow.