Every education conference now features at least one keynote about artificial intelligence fundamentally transforming how students learn. The narrative is consistent: AI tutors are coming, they will personalize education at scale, and institutions that resist this shift will be left behind.
This framing deserves pushback.
Don't misunderstand. AI has genuine applications in education. Automated grading, accessibility tools, and administrative assistance all represent legitimate uses. But the specific claim that AI will become the primary mechanism for knowledge transfer and skill development rests on assumptions that deserve examination rather than acceptance.
The "inevitability" framing is the real problem here. When something is presented as inevitable, scrutiny becomes seen as obstruction. Skeptics aren't debating evidence; they're supposedly standing against progress. This rhetorical move conveniently sidesteps harder questions.
Consider what we actually know about learning. Decades of cognitive science research points to several consistent findings. Struggle with material matters. Feedback from knowledgeable humans influences not just what students learn but how they value learning itself. Social contexts shape motivation in ways that still aren't fully understood. Relationships with teachers predict educational outcomes in ways that transcend content delivery.
None of this is incompatible with technology. But it does suggest that optimizing purely for personalized content delivery, which is what current AI tutoring systems prioritize, may address only a portion of what effective education requires.
The scaling argument deserves particular scrutiny. Yes, AI can deliver customized instruction to millions simultaneously. But scaling itself isn't educationally neutral. What gets scaled easily is often what's easiest to measure and automate: vocabulary acquisition, procedural problem-solving, test preparation. What doesn't scale as neatly? Critical thinking development, ethical reasoning, creative collaboration, or the kind of intellectual risk-taking that happens when a student feels safe enough to admit confusion to another human.
We're being asked to assume that because something can be scaled, it should be. History suggests caution here. Educational technology has a track record of overpromising. Interactive whiteboards would transform teaching. Educational software would personalize learning at the classroom level. MOOCs would democratize elite education. Each innovation arrived with similar inevitability rhetoric. Each encountered real-world friction that wasn't captured in the initial framing.
This isn't an argument for rejecting AI in education. It's an argument for rejecting the particular marketing narrative that presents AI tutoring as the future educators should accept rather than debate.
What would actual skepticism look like? It would ask: Which specific learning outcomes does AI tutoring demonstrably improve, and for which student populations? What evidence shows that AI relationships provide motivational and emotional benefits comparable to human relationships? How do we account for equity issues when deploying systems that require consistent technology access? What do we lose when we optimize education systems around what's algorithmically convenient?
These aren't anti-technology questions. They're pro-evidence questions.
The education sector should absolutely experiment with AI tools. But experimentation requires humility, not inevitability rhetoric. It requires comparing outcomes, not assuming superiority. It requires asking what might be lost, not just what might be gained.
The people selling AI tutoring as the future of education are often intelligent, well-intentioned, and convinced of their position. That doesn't make their conviction correct. And it definitely doesn't make skepticism about their claims an obstruction to progress. Sometimes skepticism is exactly what progress requires.