Abundant AI and the Shirky Principle
Since October of 2023, I’ve been giving a talk on “Artificial Intelligence and the Future of Accreditation.” Every time, one particular slide provokes a reaction.
That slide quotes Clay Shirky, NYU’s Vice Provost for AI & Technology in Education, who once said, “Institutions will try to preserve the problem to which they are the solution.”
In each audience, someone has asked, “What do you mean by that?”
Here are some examples of how schools run the risk of using AI to preserve the problem to which outdated, antiquated, or even broken models of school are the solution.
When genAI shapes interactions so that they mirror the existing Math classroom—and then gives the least and the most capable Math students the same kind of feedback—it’s helping to preserve the problem of Math as a series of disembodied, decontextualized, depersonalized equations.
When genAI gives teachers a button to generate endless “worksheet” style assignments, it’s helping to preserve the problem of assignments as question-and-answer, fill-in-the-blank, multiple choice, etc.
When genAI leads teachers to believe that they can catch kids using genAI to write their essays, it’s helping to preserve the problem of writing assignments that output a grade but that fail to activate a student’s agency or intrinsic motivation.
To combat the reality of the Shirky Principle, we need to ask, “What does great learning look like, and am I using AI to make it happen? Or am I using AI to preserve the problem—ie, ‘school’—to which I am the solution?” As Will Richardson has said many times, education and learning are not necessarily the same thing.
Combatting the reality of the Shirky Principle is a monumental task. I have deep empathy for classroom teachers, whose jobs often require them to “enter a certain number of grades in the LMS gradebook” or “cover a certain number of lessons before the end of the quarter.”
As a full time teacher, I used to divide the trimester to spot potential due dates for the three essays I would need to assign each trimester. It took me years to figure out that even if my department’s policies required a certain number of assessments, I could still get creative with the learning experience. But it’s easy for full time teachers to forget about great learning experiences in the face of endless demands. (If you have never been a classroom teacher, this post is definitely not for you!)
Still, in this world of increasingly abundant AI, we need to refuse to start the conversation with “How do we integrate AI into school?” That is a surefire way to preserve the problem to which an outdated version of school is the solution.
Instead, we need to start every conversation with, “What does great learning look like? And how can we use AI to foster that?”
As former school leaders, many of us at Middle States have devoted our careers to creating the conditions for great learning experiences. That’s why we built RAIL (Responsible AI in Learning)—to engender that conversation.
Because the Shirky Principle is real.
So ask yourself: Will I preserve the problem? Or will I obsolete it?
Christian Talbot
President
Middle States Association
Commissions on Elementary and Secondary Schools
