For generations, a large part of school has followed one pattern: learn the material, remember it, and show you remember it on a test. In a world where AI can retrieve facts, draft an essay and run a calculation in seconds, it is fair to ask whether recall should still sit at the center of that pattern.
The answer is not that knowledge stopped mattering. It is that what we build learning around needs to shift.
Knowledge Still Matters
It is worth being precise, because this argument is easy to overstate. You cannot think well about something you know nothing about. A student who knows their multiplication facts has attention free for the actual problem; a reader with broad background knowledge understands more of what they read; a doctor who knows anatomy cold can spot what a scan misses. Good teachers know this, and much of what looks like “memorization” in a classroom is building exactly that foundation.
The problem is not memory. It is a system whose incentives push recall to the center: tests that reward remembering under time pressure because recall is cheap to measure at scale, report cards that summarize those tests, and schedules that leave little time for anything harder to grade.
What AI Changes
AI and automation are absorbing a growing share of routine, predictable work, including white-collar work once considered safe:
- Legal work: AI tools can review and summarize contracts and documents, work that once filled junior associates’ days.
- Software: AI-assisted tools can draft, explain and debug code, changing what entry-level developers spend their time on.
- Writing and media: AI can draft articles, marketing copy and scripts in seconds, raising the value of judgment, editing and original reporting.
- Finance and analysis: Routine reporting and data summaries are increasingly automated.
- Healthcare: AI-assisted image analysis is being used to help clinicians detect some conditions, with people still responsible for the diagnosis.
The pattern is consistent: the tasks being automated are the ones that consist mostly of retrieving and reproducing known answers. The work that grows is deciding what question to ask, judging whether an answer is right, and applying it to a situation nobody has seen before.
What Belongs at the Center Instead
- 🚀 Reasoning with knowledge: using what you know to analyze, explain and argue, not just to recall
- 🚀 Checking answers: knowing enough to tell when a tool, a source or a classmate is wrong
- 🚀 Asking good questions: AI can answer almost anything; the skill is knowing what to ask
- 🚀 Learning how to learn: picking up a new field quickly and on your own
- 🚀 Working alongside AI: using it as a tool honestly, and knowing its limits
- 🚀 Communication and collaboration: the parts of work that are most human
This is a shift in emphasis, not a rejection of the basics. The basics become the foundation for these skills rather than the finish line.
Why the Center Is Slow to Move
The reasons are structural, not personal:
- Curricula move on a multi-year clock. Standards revisions and textbook adoptions take years, while AI changes month to month. (More on that in How Fast Can a Curriculum Change?)
- Assessment drives instruction. As long as high-stakes tests reward recall, classroom time will follow the tests. That is a design problem in the accountability system, not a choice teachers are free to make.
- Teachers are asked to adapt without the time or support to do it. Most teachers trained before generative AI existed, and most get only a few professional development days a year. Many are learning these tools on their own time anyway, because they can see what their students need.
- Credentials still certify coverage. Diplomas and transcripts record which courses were completed, not what a student can do, which keeps the whole system anchored to covering material.
What Learners and Teachers Can Do Now
- Keep building knowledge, and use it. Learn the fundamentals, then practice applying them to problems without an answer key.
- Learn with AI, not around it. Use AI tools to explain, quiz and give feedback, and check what they tell you.
- Build proof, not just transcripts. Projects, portfolios and demonstrated mastery show what you can do.
- Practice judgment. The most durable skill is deciding whether an answer is good enough to act on.
For teachers, this is where tools can take recall off your plate and give you time back for the reasoning. With DIY Degree, a student can look up the exact topic they are on, including by standard code, learn it with an AI guide that walks them through the best free explanation, and prove it with a short check that asks them to apply the idea, not just repeat it. All of that is live today. In the classroom pilot, built but not yet switched on in production, you would see only whether each student has mastered, is working on or is stuck on the topics you assigned, never their scores or their conversations, so class time can go to the discussion and problem-solving that no tool replaces.
Final Thoughts
Memory is not the enemy of thinking; it is the raw material. But in an AI age, building school around recall means building it around the one thing machines do best. The better center is what people do best: using what they know to judge, create and solve. Teachers have been pushing in that direction for years. The system can catch up with them.