Personalization at Scale: The True Potential of AI in Education

[MarisaBluestone]

This year's EconEd focused on one key idea: the decisions we make in our lives, classrooms, and society have far-reaching impacts. To explore these critical choices, we invited leading economists (who just happen to be Macmillan Learning authors) to share their perspectives on issues vital to our economy. One of the key speakers was none other than celebrated economist Justin Wolfers, Professor of Economics and Public Policy at the University of Michigan and co-author of Principles of Economics. (You can watch the full presentation here.)

There’s no getting around it—AI is the biggest elephant in the room when we talk about the future of education. In his presentation, spoke about its impact in today’s classroom. We all know how this goes. New technologies come in and change the game—some jobs or practices become obsolete, but at the same time, new and often better ways of doing things emerge. AI is no different. It’s already transforming the way we teach and learn, especially when it comes to the idea of personalization at scale.

This is true whether we’re talking about the workforce or the classroom. Imagine if every student could get a customized learning experience without needing an army of instructors. With AI, that idea is no longer a fantasy—it’s quickly becoming reality. So … how would your classroom change if you had a teaching assistant who never slept, worked 24/7, and tailored lessons to each student?

Goodbye, High-Stakes At-Home Exams

One of the first things Justin Wolfers mentioned was how high-stakes, at-home exams are “basically dead”. He noted that one of the first blows was the pandemic, which sped up the decline of traditional assessments. For many educators, the biggest takeaway from COVID was how to manage online exams—and with AI that’s being rethought again. That’s because AI tools like ChatGPT have made it easier than ever for students to cheat. 

So, what’s the solution? Well, instead of worrying about how AI is making cheating easier, Wolfers believes that we should be thinking about how to use it to our advantage. He noted that AI can help us develop new ways of assessing student learning, ways that are more meaningful and less vulnerable to dishonesty.

Adrian Wooldridge, who used to be the editor of The Economist, suggested that we go back to sitting in a room with a tutor for an hour each week. “The problem with that idea is that it’s remarkably uneconomic. I teach 500 students. The idea that we could hire 499 other colleagues to help is unrealistic,” Wolfers said. That’s where AI can be a game-changer. 

With AI, however, instructors can offer each student a personalized set of AI tools, providing both scale and personalization—at a fraction of the cost. And that’s the big idea: personalization at scale. Wolfers shared four ideas for applications where this could benefit teaching and learning: a Socratic tutor, a practice exam coach, a teaching collaborator, and a text aggregator.

Hello, AI as a Teaching Assistant

Wolfers first mentioned the idea of a Socratic style tutor—an AI tool that doesn’t just hand over the answers but instead asks guiding questions to help students think critically and learn. He cited a study out of Turkey that found that students who used a GPT-powered tutor outperformed their peers who were using more traditional tools. He believes tools that don’t just make learning easier but make it more effective is what’s needed the most. “And guess what?,” he said. “A version of this Socratic tutor already exists.” 

Available in Macmillan Learning’s Achieve, “it’s like having a teaching assistant that’s always ready to help” by guiding students through tough material without giving away the answers. It’s multilingual and can seamlessly switch languages to help non-native speakers. You can get a free demo of the tutor by clicking here.

An AI That Encourages, Summarizes and Aggregates

Wolfers also introduced the idea of AI acting as a practice exam coach. To understand what it could do, picture this: a coach that gives you practice questions, provides hints, and offers feedback all while cheering you on. He noted that it would not just be practicing for the sake of it; rather, it would give students the freedom to practice until they feel ready.

In his class, he offers help in a fun and engaging way with a coach he created that blended Ted Lasso’s sunny disposition with economic knowledge. The coach was designed to give practice questions, provide hints, and offer feedback in an encouraging way, allowing students to practice as much as they want until they feel ready. He noted that his personalization could be a total game-changer, especially for students who need a little extra support before exams.

Stop, Collaborate and Listen 

Wolfers suggests that GPT could be used to generate active learning activities for a topic (like externalities in his economics class.) However, the key to its effectiveness is to work hard at making it work well. 

In this case, a simple prompt may not give the results you’re looking for. It’s the difference between a bad prompt like, “Give me a class activity to explain what externalities are” and a good prompt like, “Create an interactive classroom activity for my freshman principles of economics class where students identify positive and negative externalities from everyday situations, explaining how these affect different stakeholders and proposing solutions to mitigate the negative impacts  Ensure the activity involves small group discussions where students brainstorm examples, categorize them, and present their findings.”

Simplify Student Feedback

Now that you’ve stopped to collaborate and listen, this could be a whole new invention to supplement what Wolfers calls “one of the most tedious parts of teaching” -- sifting through tons of student feedback. Wolfers described how AI can summarize unstructured text—like student feedback—in seconds. This not only saves time but gives us a real-time snapshot of what students are struggling with, so we can address those issues in the next class.

“It’s incredibly useful in deciding where to go next in a lecture or for gaining insight into what students are struggling with," he said. Indeed, using AI to help teach could have broad applications, from class feedback to improving teaching methods in large classes

Ultimately, the true power of AI lies in its ability to personalize learning at scale. From Socratic tutors and practice exam coaches to teaching collaborators and feedback summarizers, AI is opening up possibilities we couldn’t have imagined just a few years ago.  So are we ready to leverage AI to its fullest potential and meet the future head-on? 

Sure, AI comes with its challenges—cheating, over-reliance, and the like—but the opportunities far outweigh the risks. As Wolfers put it, AI is our chance to personalize education in ways that were never possible before. It’s time we embrace that potential, figure out how to make these tools work for us, and ultimately help our students succeed in this new era of education.  Watch Wolfers full presentation from EconEd 2024.

Want more EconEd? Read: Spend Billions to Save Trillions: An Economist’s View of COVID-19 Response at EconEd as Alex Tabarrok, co-author of Marginal Revolution and Macmillan Learning’s Modern Principles of Economics shares “The Economic Way of Thinking During a Pandemic.”