AI Can Teach Students but Only Teachers Can Force Them to Learn

AI is better at one-on-one tutoring than most teachers. It is highly knowledgeable, precisely customizable to the student’s needs, adapts as the student learns, and is infinitely patient. There’s research proving this. And yet, AI will make little difference to a majority of the students. Because AI tutoring only works if the student wants to learn and engages with the AI—something only the top 5% of students can do. For the rest, AI can’t make the student want to learn: only a teacher can provide the carrot and stick necessary for this to happen.

I made this point in AI in EdTech Should Help Teachers Because It Can't Replace Them — (Lessons from the failure of Khanmigo, Khan Academy's AI-powered Tutor) but, here’s another article, this time by an experienced teacher, making the same point more forcefully and with more details. Here are some excerpts.

The problem isn’t specific to AI; it afflicts all ed-tech. We’ve seen this play out with online courses, Coursera, and Khan Academy’s videos, and everyone getting excited with flipped classrooms—the idea that students would watch intructional videos at home and come to class for asking questions and clearing doubts. Anyone who’s been a teacher knows that this never works, because the students don’t watch the videos at home:

For students who are self-disciplined and acquire metacognitive skills, Khan Academy and flipped classrooms can work. However, the majority of students do not fall into this category. Most lack the skills and discipline required to pay attention, stay focused, and identify what they don’t understand.

This may be the reason why the flipped classroom model did not really catch on. People assumed that students would watch video lessons at home and come to class prepared for practice. The reality was that many students watched them passively or did not watch them at all. It became such a widespread problem that proponents of the flipped classroom wrote articles to answer the question, “What if students don’t watch the video?”

And the same problem is repeating itself with AI tutors:

I am not questioning whether AI adaptive learning platforms can explain concepts better than a teacher. In some cases, it probably can. Especially when learning experts are behind the design. The real question is whether AI can reliably sustain the attention, effort, persistence, and metacognitive monitoring required for learning.

Here is the core of the argument:

Meaningful learning often requires effort. It requires students to make mistakes and work through confusion. For students who are highly motivated and have strong metacognitive skills, this challenge can be manageable. However, most students naturally try to avoid this discomfort and go towards the path of least resistance. For example, a student may copy from a partner or rely on AI to generate responses without actually thinking through the concepts.

In a traditional classroom, that is where teachers play an important role. Teachers provide instruction AND the accountability students need to stay engaged in the learning process. This can include checks for listening, choral responses, mini whiteboard checks for understanding, cold calling, Perch’s stance, etc. These practices make it harder for students to disengage without anyone noticing.

In contrast, AI adaptive learning platforms cannot fully replicate the human presence that creates accountability in a classroom. It cannot tell when a student is mentally checked out, staring at a screen without processing information, or clicking through tasks with minimal effort.

If you are a teacher or are interested in the impact of AI on education, you should definitely read the whole article; it has a lot of detail, and answers to frequently asked questions (like, “If students are using AI adaptive learning platforms, should the teacher’s role shift toward helping students monitor their progress and respond productively when they get stuck?” … the answer is No, but you’ll have to read the article to find out why.

In short, teachers remain necessary and important, to provide the mentorship and inspiration (the carrot) and the structure and discipline (the stick), without which, AI and other ed-tech cannot help the students who don’t help themselves (which is 95% of them).