AI Student Spectators

States are trying to figure out how to respond to AI in schools, and they are most flubbing it. A piece from CT Insider shows just how far in the weeds folks are getting.

The piece by no less than five staff writers (Natasha Sokoloff, Crystal Elescano, Ignacio Laguarda, Jessica Simms, Michael Gagne) looks at how Connecticut’s district approaches are working out in the classroom, and the items touted as success are... well, discouraging. Meanwhile, the state is putzing along and “plans to build its formal AI guidance for all districts based on the findings of the pilot program; collaboration with experts and AI educational organizations; and research-based documents ‘to ensure we get this right,’ [state academic chief Irene] Parisi said.”

Westport Public Schools has AI tools in place that are, according to Parisi “education-specific and have privacy protections.”

“They said it was like having a teacher in their pocket,” she said. The tools could help students work through a particular problem, brainstorm ideas, research for projects and provide feedback, she said.

“Help” and “work through” are doing some heavy lifting here. “Provide feedback” remains one of the popular items in the AI arsenal. I remain unconvinced. Feedback that does not understand or include student intent-- what they thought they were doing, what they meant to do-- is just correction. “Do this instead of that.” If you don’t know why the student did “that” in the first place, you can’t provide much in the way of useful correction, and since AI does not “know” anything, all it can do is edit the student’s work for them. What do students learn from this? This is the pedagogical equivalent of an adult who shoulders the student aside and fixes their work while the student watches.

But the proud example of an AI project, shared by the superintendent in a board meeting, is even worse.

Students in a middle school social studies class used AI to create and question “digital peers” and “characters” from the Middle Ages while the teacher guided them in evaluating responses for accuracy and evidence.

Many teachers (including me) would recognize this assignment immediately, only Back In The Day, we would have the students create and role play the characters themselves. In Mrs. O’Keefe’s eighth grade English class (back in 1971), we had to research a historical person and then portray them as a guest on a talk show (my friends Andy and Stewart drew Van Gogh, and in the middle of his interview he became over-emotional and cut off his own ear, complete with fake blood). My sister-in-teaching Merrill annually had her students put Milton’s Paradise Lost on trial, with students role playing characters from the work.

This is a variation on that same assignment except AI does the role playing and students are transformed from actors into spectators.

Almost any version of this assignment would be better. Let students role play. Let them craft faux social media accounts for their characters. Anything that had them actively creating the character based on their own research, rather than feeding some stuff into an AI and sitting back to observe and judge the result. What does the teacher even assess in such an assignment? How is this any better than just watching a video about the topic?

If you’re considering incorporating AI in your lesson and wondering how to decide what to have it do, here’s a hint-- do not have it turn students from active participants into spectators who simply watch what the bot does for them. Students should be main characters in their own education, and not observers, sidelined so that the plagiarism machine can shine.