AI Won’t Transform Education Unless Schools Change

I wrote this article about two years ago and here is a summary.

Artificial intelligence has arrived in education with extraordinary speed. Since the release of ChatGPT, schools have been flooded with conferences, workshops, books, and advice about how teachers and students should use AI.

But much of this conversation is too small.

The central question is not simply how teachers can use AI to improve lesson planning, feedback, tutoring, or assessment. The more important question is:

What if schools themselves need to change before AI can make a real difference?

If AI is inserted into traditional classrooms to teach the same subjects, meet the same standards, and produce the same outcomes, it is unlikely to transform education. It may make some tasks easier or faster, but it will remain an add-on to an old system.

Personalized Learning Is Not Really Personalized

One of the most common promises of AI is personalized learning.

But most forms of personalized learning are not truly personal.

Students may work at different speeds. They may choose different places or formats for learning. They may receive different explanations. Yet they are still expected to learn the same content and achieve the same outcomes.

That is not personalization in any meaningful sense.

Real personalized learning should begin with the student as a person—with individual strengths, interests, experiences, talents, and passions. Students should not simply receive personalized instruction. They should increasingly personalize their own learning.

AI can help make this possible. It can support students as they explore interests, acquire knowledge when they need it, pursue ideas, and develop areas of unique strength.

From Assigned Projects to Problems Worth Solving

Project-based learning also needs to change.

Too often, the “project” is predetermined by teachers. Students are given a problem and then asked to complete the process. The emphasis is frequently on participation rather than on whether the final product is actually useful or valuable.

In the age of AI, students can do something much more meaningful.

They can find and solve problems themselves.

That means learning how to notice problems, decide which ones are worth pursuing, refine them, seek feedback, improve ideas, and create solutions that matter.

Those solutions could take many forms: music, art, writing, tools, services, designs, inventions, or something entirely new.

AI can support students throughout this process, but the student remains responsible for identifying what is worth doing.

Schooling Itself Must Be Redesigned

Once personalized learning and problem finding become central, the traditional organization of schooling begins to look increasingly restrictive.

A rigid curriculum divided into subjects, grades, and age groups makes genuine personalization difficult.

A different curriculum could include three broad components:

learning necessary for participation in society and citizenship;
foundational knowledge and skills communities consider essential, such as literacy, numeracy, digital literacy, and financial literacy;
learning personalized by students themselves.

This would preserve a common foundation while creating substantial space for students to develop their own interests and strengths.

Teachers, Assessment, and Grouping Must Change Too

If learning changes, teaching must change.

Teachers do not need to remain primarily instructors delivering content to groups of same-age students. AI tools, online resources, peers, and outside experts can all support learning.

Teachers can increasingly become coaches, mentors, advisers, and designers of learning experiences.

Assessment must also change. If students pursue different strengths and solve different problems, standardized tests become less useful. Assessment should instead document students’ growth, interests, accomplishments, problem-finding ability, and quality of solutions.

Even age-based grouping can be reconsidered. Students could be organized around shared interests, problems, or learning needs rather than simply by birth year.

Big Change Can Start Small

The problem, of course, is that schools are difficult to transform.

One practical approach is to create a “school within a school.”

Instead of asking an entire school to change at once, willing teachers and students can create a different learning environment inside the existing institution.

This smaller space can experiment with personalized learning, problem finding, new forms of assessment, flexible grouping, and extensive use of AI. If it succeeds, others can observe it, join it, and adapt its practices.

Transformation can therefore begin without requiring immediate system-wide agreement.