What is Worth Knowing in a World of AI?

Ask a room full of educators what students need to know and you might get a hundred different answers. Ask your favorite AI app the same question and you will get an answer in three seconds, footnoted, formatted, and wrong in ways that are hard to spot.

That is the actual crisis we are dealing with. We never finished deciding what was worth knowing in the first place, and now we have handed the question to a machine that will answer it whether we are ready or not.

Grant Wiggins and Jay McTighe saw this coming 25 years before ChatGPT existed. They called it coverage. It seemingly matters more today than ever.

The Three Circles

In Understanding by Design, Wiggins and McTighe gave teachers a simple but brutal filtering tool, laid out in a short ASCD framework overview worth reading in full. Picture three concentric circles. The outer ring holds everything worth being familiar with, the broad exposure stuff, names and dates and vocabulary that build context. The middle ring holds knowledge and skills that are important to know and do, the working knowledge a discipline actually runs on. The small circle in the center holds enduring understandings, the ideas that outlast the test, the unit, the school year. The stuff a person still uses at forty even if they could not pass the quiz on it today.

Most curriculum lives in the outer ring because the outer ring is easy to teach and easy to test. Coverage is comfortable while we know that depth is not.

The problem AI just exposed is that everything in that outer ring (the facts, the definitions, the dates, the surface-level recall) is now instantly retrievable.

A student does not need to hold it in memory when a model can produce it in a sentence. So if we keep building classrooms around the outer ring, we are training kids to compete with a tool that will always beat them at that game.

The center circle, the enduring understandings, is the only ring AI cannot do for a student. It cannot want to understand something. It cannot wrestle with an idea until it becomes part of how a kid sees the world. That work still belongs to the learner which means Wiggins and McTighe were not just describing good curriculum design, they were describing the only part of school that survives contact with AI.

What Darling-Hammond Adds

Linda Darling-Hammond's research fills in the piece Wiggins and McTighe leave open, which is how you actually get students into that center circle instead of just writing it into a unit plan.

In Implications for Educational Practice of the Science of Learning and Development, Darling-Hammond and her co-authors synthesized decades of research across the learning sciences and concluded that transferable understanding depends on strong relationships, sustained practice, and learning environments built around real application, not one-off exposure to content. That is the delivery mechanism for the center circle. You do not get there through more coverage. You get there through repeated meaningful practice with ideas, knowledge, and concepts.

AI could go either direction here. It could become the tool that finally makes deep, performance based instruction affordable and scalable for every school, not just the ones with the budget for small class sizes and coaches. Or it could become the fastest, cheapest way yet invented to keep kids parked in the outer ring, producing polished outputs that look like understanding but never required any.

Which one happens will be a design choice teachers and school leaders make every day.

Where Learning 3.0 Comes In

This is the exact fork in the road Learning 3.0 is built around. AI in education has to serve learner agency, not algorithmic efficiency. Efficiency wants the outer ring. Efficiency wants fast answers, clean outputs, coverage at scale. Agency instead wants the center ring. Agency wants a student who can say why an idea matters, defend a claim, revise thinking after being wrong.

Every time a tool makes it easier for a student to skip the struggle instead of scaffold the struggle, you have let efficiency win. Every time a tool clears away the outer ring busywork so a student has more time and energy for the center ring work, essential questions, transfer tasks, performance defense, you have let agency win.

That is the actual test for whether an AI tool belongs in any learning environment. Does it save time on the outer ring so kids can spend more of it in the center?

We also can’t forget how important that second ring is. Students need to build knowledge and skills in order to get to the enduring understanding and transfer. It does not happen by magic, but instead through deliberate practice that meaningful and relevant. To skip the second circle, you’ll never get to the deep learning.

What This Looks Like Monday Morning

Run your next unit through the three circles before you plan a single lesson. Be honest about what is genuinely worth being familiar with, what is important to know and do, and what is the one or two enduring understandings the whole unit exists to build. If you cannot name the center circle idea in one sentence, you probably have not finished designing the unit yet, or are missing key understanding as the instructor.

Then ask where AI fits. Maybe it can handle outer ring work. Vocabulary building, background research, first draft generation, practice problems. Protect the center ring completely. That is where kids write in their own voice, defend their thinking out loud, revise after critique, and connect the idea to something in their own life. No shortcut belongs there.

Darling-Hammond's research gives you the delivery mechanism after knowledge is built. Lean into performance tasks, authentic work, and iterative revision. Wiggins and McTighe give you the filter for what deserves that kind of attention, while Learning 3.0 gives you the compass for where AI helps and where it quietly steals the exact struggle a student needed.

What is worth knowing in a world of AI was never really a new question. Wiggins and McTighe were asking it in the nineties. Darling-Hammond has spent a career proving which instructional moves actually get students there. The schools that treat this as a design problem, not an AI or tech problem, are going to be the ones whose students come out the other side actually able to think.