Back in 2007 Jon Bergmann and Aaron Sams pioneered what is now known as flipped learning. Today, Bergmann, who remains a high school science teacher in addition to being a recognized education expert, says a new teaching paradigm is underway because of the emergence of AI.
To that, Bergmann has developed a new teaching framework that combines flipped learning and mastery learning models with the modern realities of AI. He calls it The Mastery Flip, and its three pillars are designed to maximize the beneficial impact of AI while minimizing its harms and preparing students to harness this technology for the future.
Pillar 1: AI Engines (Turbocharging the Independent Space)
AI can help students when they’re working at home in what flipped educators often refer to as the independent space. Ultimately, Bergmann would like to see teacher-controlled AI tutors that are trained to ask students questions rather than provide answers, and can be programmed to reflect a teacher’s voice, style, and preferences.
“In the world of chemistry, there are two camps in how you do the math part of chemistry,” Bergmann says, advocating for the option of teachers choosing which method for the AI to emphasize. “I want it to be tuned to the teacher.”
Many AI tools aren’t quite there yet, but are getting close. Bergmann is optimistic that tools that provide more teacher control, along with rigorous data and privacy control for students, are on the way.
One tool he’s impressed with is Skylow.AI, which allows users to upload video and audio of themselves to create an AI clone. So now when Bergmann’s students watch one of his flipped learning videos, they can raise their hand and ask questions about the material, and an AI version of Bergmann will answer.
He reports that the tool has so far split the class, with half liking it but the other half saying, “It’s too creepy that there’s an AI Mr. Bergmann.”
Fear of robot clones aside, the point remains that the right type of AI technology could be really beneficial to students who are working outside of the classroom.
Pillar 2: Analog Roots (The Anchor of Learning)
The availability of AI and the way it can be used to generate student work makes the teacher’s work with students during class time even more important than ever before. That’s why Bergmann’s current approach is to encourage more tech outside of class to support learning but less during class so students can focus on cognitive work and learning from the teacher.
“My class is becoming less and less digital,” he says. Though he adds, he can imagine a role for well-designed AI tutors even in class, provided the right oversight is in place and the AI supports a student’s productive struggle rather than giving them all the answers.
“The LLMs were designed for businesses, and for efficiency, and not for education, and productive struggle is not built into them, so we need to have something that’s built for kids and their brains,” Bergmann says.
With or without the help of AI, the idea is to use class time to ensure students themselves are working, struggling, thinking, and ultimately making progress in their learning.
Pillar 3: Human Checks (Validating the Cognitive Journey)
Many teachers seem unaware of the way in which AI has fundamentally changed what effective assessment should look like, but Bergmann does not have any such illusions. Traditional research papers, for instance, no longer work. “You can’t do that anymore,” he says. “The game’s over. In two seconds, I can prompt ChatGPT to write the best research paper.”
Because of this, teachers need to create assessments that ensure students, and not some AI chatbot, have really worked through the material and thought about it.
For example, in one assignment for Bergmann’s physics class, students have to build a trebuchet, a complex medieval catapult-type weapon. The final for the assignment involves students using these beefed-up catapults on the football field to see how far projectiles can be launched. “Then they have to explain the physics behind it, but I’m smart enough to know that they’re probably going to have AI write a lot of that, so I’m going to just accept that,” he says. But it’s only going to count for a portion of their grade; larger portions will be based upon the performance of the catapult and the student’s ability to explain the physics behind it to Bergmann. “We’re going to have a marker board or a whiteboard, and I’ll say, ‘Explain how this worked. Why did this happen?’”
These types of human checks can ensure that students are not outsourcing the knowledge and critical thinking portion of assignments to AI, which Bergmann worries may already be happening too much.
“My terrible prediction is I think it’s going to be very bifurcated. Those students who use AI as a cheat code are going to really struggle in life,” he says. “This is an electricity-level event right now that’s happening in our lives. The vast majority of candlemakers went out of business when electricity came along.”
Bergmann hopes his new model can help educators guide students to avoid the fate of candlemakers and use AI to supercharge learning.