A growing number of students are no longer using AI only to “get answers.” They are using Large Language Models as tutors, study partners, exam coaches, writing companions, and conceptual sparring partners.

A typical prompt now looks like this:

“You are a neuro-optimized tutor. I want to learn a complex skill faster. Create a weekly learning blueprint based on spaced repetition, interleaving, the Feynman technique, and active recall. Apply it to [topic]. I want to reach a high level in 90 days.” source

Behind the exaggerated language, something important is happening.

Students are discovering that an LLM can explain a concept ten different ways, generate examples, ask questions, correct misunderstandings, create quizzes, simulate debates, and adapt instantly to their level of confusion. In many cases, this feels more personal, more interactive, and more available than a traditional class.

Is learning with LLMs becoming more effective than learning in class?

The answer is probably not “yes” or “no.” The better question may be:

What kind of learning are we talking about?

For conceptual understanding, LLMs can be extraordinarily powerful. They allow learners to move quickly between explanation, questioning, application, correction, and self-testing. A student can ask: “Explain this like I am 12,” then “Now explain it technically,” then “Give me three counterexamples,” then “Quiz me until I understand.” This kind of active, adaptive dialogue is difficult to reproduce in a large classroom.

But classrooms offer something different yet essential: shared attention, human judgment, social commitment, peer discussion, mentorship, embodied presence, and the slow formation of an intellectual community. A good classroom is not just a place where information is delivered. It is a place where meaning is negotiated, challenged, and made public.

Perhaps the real opportunity is not to choose between LLMs and classrooms, but to redesign their relationship. LLMs may be excellent for personal conceptual iteration. Teachers may be essential for intellectual orientation, critique, community, and standards of truth. Classrooms may become less about delivering explanations and more about organizing inquiry, dialogue, practice, and shared meaning.

We would love to hear from teachers and professors:

Where do you see LLMs genuinely improving learning?

Where do they create false confidence or shallow understanding?

What parts of teaching should never be automated?

How should assignments, exams, and classroom discussion change?

What would a good “LLM-native” course look like?

How can teachers guide students to use these tools responsibly and deeply?