Schools Test Answers. The World Needs Better Questions. Time to Rethink Education.

This thought has been nagging me for a while. I chat with my kids about school, about how they learn, about what they actually use. Then I thought, why don't I write this down and see what everyone thinks.

84% of high school students use generative AI for schoolwork (College Board, 2025). 92% of UK university students use AI tools (Higher Education Policy Institute, 2025). They're using it whether we allow it or not.

So why are we still grading essays that a machine can write in 10 seconds?

The current system tests recall. Memorize facts, repeat on a test, get a grade. That made sense when the teacher was the only source of information and books were expensive. Information was scarce. Today it's free and instant.

What's scarce now is knowing what questions to ask, how to evaluate whether an answer is correct, and how to apply knowledge to real problems. These are the skills that matter at work. These are what we should be teaching.

Feynman called this out decades ago. He reviewed textbooks for California schools and found what he called "cargo cult education." Students going through the motions of learning, reading, highlighting, memorizing, without actually understanding anything. The form of education without the substance. Books teaching students to repeat words without grasping the meaning behind them.

That was 1964. Same problem exists today, just with fancier tools.

Here's what I think should change:

1. Teach fundamentals deeply. Math, logic, writing, critical thinking, communication. These don't change with technology. A student who understands first principles can learn any tool.

2. Let students use AI openly. Stop pretending they're not using it. Teach them how to use it well. How to verify answers. How to ask better questions. How to spot when the tool is wrong. These are real skills they'll need at work.

3. Evaluate thinking, not output. Grade students on their reasoning, their ability to defend their solution, their ability to identify what they don't know. Not on whether they produced a 5-paragraph essay that a machine could write.

4. Give real problems. Not textbook exercises with known answers. Real messy problems where the student has to figure out what information they need, find it, evaluate it, and build something.

And while we're at it, why do we still have certification exams that test multiple choice recall? AI can answer those questions more accurately in seconds. Many people pass these exams by learning to eliminate wrong answers, not by understanding the material. That's not where real knowledge lives. Real knowledge is applying concepts to a situation you've never seen before, explaining your reasoning, and knowing when something doesn't apply.

When these students enter the workforce, they'll be encouraged to use AI tools. Every company is adopting them. We're training students to hide their tool usage, then expecting them to embrace it at work. Makes no sense.

Some institutions are starting to move. There are executive orders promoting AI in classrooms, student AI bills of rights being drafted, universities redesigning assessments. But progress is very slow. Most are still fighting it instead of adapting.

The question isn't whether AI will change education. It already has. The question is whether we redesign the system around reality or keep pretending it's 1995.

What do you think? Should we teach students how to use these tools effectively, or keep banning them and testing memorization?

Comment and let me know what you think.