How Should You Use AI this School Year?

What are the current limits of AI and how does that work in an educational setting?

"It'll take less than five minutes,” Chat confidently assured me after I asked it to create a mnemonic device to help me memorize John Donne's sonnet When I Consider How My Light Is Spent. Earlier this year, I shared that one of my New Year's resolutions was to use AI every day, and on this particular afternoon, I thought I'd found a productive use for it. My hobby is memorizing poetry and I’ve gotten Walt Whitman’s A Promise to California and Donne’s Death Be Not Proud down pat after several weeks on each. I figured AI might help me speed up the process. Gentle reader, this was a bald-faced lie.

Five minutes did not suffice to memorize 14 lines of archaic English. As it turns out, memorizing seventeenth-century poetry still takes work!  AI was wildly overconfident in predicting how quickly I would be able to commit the poem to memory with its mnemonic tricks.

AI has promise for educational organizations. I hear hopes that artificial intelligence will become a long-suffering paraprofessional who happily spends the day formatting lesson plans, completing grant reports, preparing school board presentations, and handling the mountain of paperwork that comes with modern education. If that vision became reality tomorrow nobody would complain. But I don’t think that’s realistic. AI is a tool, not a coworker, and requires a lot of direction and data quality.

What makes AI work in education?

Like every tool we've adopted in education, when we learn to use AI well, we benefit tremendously! Imagine if we were still writing on clay tablets to practice our addition facts. However, AI isn’t as “off the shelf” as it’s often purported. 

When I first started using AI, I got frustrated daily. The responses felt generic and the writing wasn't particularly helpful. I couldn’t understand why everyone seemed so excited about a tool that wasn't saving me very much time. Looking back through those early conversations, my prompts were pretty awful. I was asking broad questions, providing almost no context, uploading no supporting documents, and then wondering why the responses lacked the specificity needed to be useful. We've all heard the old software saying, "Garbage in, garbage out." My AI chat history was a real trash day.

I changed my approach to AI prompting after I read Ethan Mollick's work on AI and implemented those helpful tips (you can read more in this Data-Driven Dispatch). I also resolved to use AI everyday and those extra practice sessions helped me understand that for AI to work well, you need to explain the assignment’s success criteria explicitly. I gained a lot of clarity when I had to work on explaining my own problems, which helped me better know what I wanted. After dozens of conversations (and plenty of disappointing results!) I found my way into using AI more confidently.

Before I begin an important conversation with AI, I now include previous reports, spreadsheets, examples of prior reports, style guides, or background information about the project. The difference in the quality of the responses is remarkable. This context has made all the difference.

How do schools use AI?

I think schools are approaching AI in much the same way many of us approached our first student information systems. We hope the technology itself will solve a problem, but technology acts as a force multiplier; it is dependent on the quality of the process. If your data is incomplete, AI won't know what you forgot to include, but it’ll be happy to invent something you don’t need. And if your organization has unique reporting requirements, AI can't possibly know them unless you teach it that. That's why data quality is so essential for using AI well. When you have a messy spreadsheet, AI is happy to organize it, but AI can’t tell you that your process for collecting information is giving you junk data. As Tyler Cowen, noted AI economist says, “For instance [with AI], I can complete many research tasks more quickly, but it does not improve the quality of the final product, which is still bounded by my talent, not the capabilities of the AI. “

If I could offer one piece of advice to a superintendent, principal, or teacher who wants to make better use of AI this year, it would simply be to use it every day for a few weeks. Better yet, find an AI buddy. Compare prompts and share your tops. The only way to understand where AI shines and where it struggles is to spend time working alongside it. AI will continue to improve. New models will be released, new capabilities will emerge, and today's best prompting strategies will eventually become outdated. What won't become outdated is the willingness to learn. Educators have always adapted to new tools, and AI is no different.

Final Thoughts and How to Take Action

How has AI surprised you? Has it become a valuable partner in your work, or are you still yelling at the computer screen? Join the conversation on social media and share your thoughts!

About Instructional Data Solutions: Our mission is to remove barriers to enable education organizations to focus on what matters most. We assist school systems and education organizations of all sizes by collecting, analyzing, and clearly communicating data. We are committed to empowering educators through comprehensive data analytics and tailored support. Additionally, we provide dedicated support for operational needs, process improvement, and special projects, offering customized solutions to enhance effectiveness and success. Our solutions bridge the gap between data collection and instructional improvement in PK-12 settings.