What mastery learning teaches us about preparing students for an AI-driven world

By Bob Hyjek, Dean of Enrollment at Stuart Hall School

For most of the last century, schools built their programs around a fairly safe bet: we could predict, with reasonable accuracy, what students would need to know and do as adults. Content changed slowly. Careers followed known paths. A school's job was to transmit a fixed body of knowledge as efficiently as possible, and grading systems were built to measure exactly that: how much of the material did a student retain and reproduce correctly.

That bet is a lot harder to make today. Artificial intelligence tools are changing how

professionals write, research, and solve problems on a timeline measured in months, not

decades. The jobs many of our current students will hold don't exist yet, and the tools they'll

use to do that work are still being invented.

This is the kind of volatility, uncertainty, complexity, and ambiguity, sometimes shortened to VUCA in leadership circles, that schools now have to prepare students for directly, not as an afterthought. At Stuart Hall, we've found that mastery-based learning, a model we adopted well before AI became a daily classroom conversation, turns out to be unusually well suited to this moment. Not because we designed it with AI in mind, but because the skill it builds, the ability to revise, iterate, and improve based on specific feedback, is the same skill students now need to work well with AI tools.

What mastery learning actually looks like

Traditional grading measures how many correct answers a student produced, out of 100, on a single attempt. It's efficient, but it tells a student very little about how to improve. A 78 doesn't say what to do differently next time.

Mastery-based learning replaces that scale with a rubric, typically four levels, that describes

what a student can and can't yet do with a specific skill. Consider a history student learning to make a defensible historical argument. Under a traditional model, that student gets a grade and moves on. Under a mastery model, the rubric tells the student precisely where the gap is: she can take a defensible position, but she hasn't yet learned to build a line of reasoning that justifies it. That's not a grade. It's a diagnosis, and it's specific enough to act on.

This plays out across disciplines in very concrete ways. History students at Stuart Hall conduct archival research at a local presidential library rather than reading about primary sources secondhand. Science students learn to select and apply the right tool for a given lab problem, not just follow a fixed procedure. In both cases, the target isn't a single correct answer. It's a skill that gets revised and strengthened over multiple attempts, with specific feedback driving each round of improvement.

Why this matters more now, not less

Here's the connection that doesn't get made often enough: working well with AI tools requires

almost exactly this same muscle. A student who has only ever been rewarded for a correct first attempt has no practice iterating toward something better. A student who has spent years refining work based on specific feedback, on what's missing and what to try next, already knows how to treat a first draft as a starting point rather than a final answer.

We see this show up discipline by discipline. In language classes, students studying rhetoric are learning the same skill that shapes an effective prompt: how word choice and structure change the response you get. Philosophy coursework, built around identifying gaps in logic, gives students the tools to spot flawed reasoning whether it comes from a classmate or a language model. In science and math, students who've practiced judging whether a solution is not just correct but efficient, or even elegant, are better equipped to evaluate AI-generated output rather than accept it uncritically. And history's emphasis on source evaluation, on asking where information came from and whether it can be trusted, is directly transferable to assessing what an AI tool produces.

None of this is a claim that mastery learning was built for AI. It wasn't. But a model built around iteration, specific feedback, and revision was always going to translate well to a world where the most valuable skill is knowing how to get better at something you don't yet understand.

A Question Worth Asking

For other schools, agents, and families evaluating academic programs right now, we'd offer this: instead of asking what a program teaches, ask how it teaches students to improve. Does feedback tell a student exactly what to do differently, or just how many points they lost? Can a student explain, in specific terms, what separates their current work from stronger work? Programs that can answer yes are building the kind of adaptability that no one can fully predict the need for, but that almost everyone will need.