AI: Good Instructor, Terrible Educator

The painting "The Predictor" by Giorgio de Chirico, created in 1919, depicts a blindfolded figure poised to throw a ball, surrounded by architectural forms that suggest both prediction and uncertainty. Nearly a century before large language models became ubiquitous in university classrooms, de Chirico captured something essential about the illusion of certainty that technology promises. Today, that same blindfolded confidence characterizes how students interact with artificial intelligence, and the consequences are far more troubling than the artist could have imagined.

My recent experiences as a university professor have made it clear that we need urgent, critical examination of how artificial intelligence tools are proliferating through higher education. AI demonstrates considerable promise as an instructional aid, but its role in fostering genuine intellectual development seems increasingly geared toward transforming eager students into dependents with atrophying learning skills. The distinction matters profoundly: we are not merely changing how students access information, but fundamentally altering how they develop the capacity to think.

The Illusion of Interactive Comprehension

Artificial intelligence excels as a perpetually available instructional companion, capable of walking you through any topic on demand. This unprecedented accessibility functions as a sort of interactive Wikipedia, representing genuine advancement in democratizing knowledge. Students can ask questions at 2 AM, receive immediate responses, and proceed with their coursework. On the surface, this appears to solve persistent problems of educational access and support.

However, the technology's limitations create a dangerous asymmetry. AI still hallucinates on a semi-regular basis, dispensing shallow, incomplete, or misleading information with complete confidence. Effective use requires maintaining robust critical thinking skills at all times—the very skills that undergraduate and freshman students are still developing. The metacognitive discipline necessary to maintain intellectual skepticism when engaging with AI-generated content doesn't come built-in; it must be cultivated through precisely the kind of difficult, sustained effort that AI makes unnecessary.

Students in their formative years frequently lack this discipline. Instead, they experience immediate gratification from answers that seem complete enough to convince peers and professors. The temptation to "prompt the 'ol LLM" and run with the output is nearly irresistible when deadlines loom and the work feels difficult. But this creates a fundamental confusion between instruction and education.

The Necessity of Productive Struggle

Authentic education, as opposed to mere information transfer, requires learners to engage with assignments deliberately designed to be difficult, painful, and tedious. This is not cruelty on the part of educators—it's the mechanism by which cognitive boundaries expand. Genuine intellectual growth emerges from successfully navigating tasks that initially appear insurmountable. This productive struggle forms new neural pathways and cultivates intellectual resilience.

Consider why learning mathematics or a musical instrument requires so much diligent practice. The difficulty is not an unfortunate byproduct; it is the essential feature. Each moment of frustration, each attempt that fails, each tedious repetition contributes to the development of genuine understanding and capability. When you struggle through a complex proof or practice scales for hours, you are not merely accumulating information—you are rewiring your brain to think differently.

AI undermines this pedagogical mechanism by its very nature. When confronted with challenging academic work, students increasingly demonstrate a reflexive tendency to delegate cognitive labor to AI systems rather than persevering through the discomfort that characterizes meaningful learning. This behavioral pattern reflects deficiencies in the diligence and discipline required for sustained intellectual effort. Over time, not only do learning skills atrophy, but the very capacity for independent thought begins to deteriorate.

The Dependency Spiral

The most urgent issue is the self-reinforcing nature of AI dependency. As students progressively outsource cognitive tasks, they become increasingly reliant on these tools, creating a dependency spiral with troubling cognitive and psychological implications. Each instance of using AI to avoid difficult work weakens the student's confidence in their ability to handle such work independently.

Eventually, when confronted with the necessity of independent work—perhaps during proctored exams, oral defenses, or professional situations—many students find themselves in a state of denial, perceiving, perhaps accurately, that their autonomous learning capacities have atrophied beyond easy recovery. This perceived point of no return creates powerful incentives for continued concealment of and dependence on AI. The student becomes trapped in a cycle where the tool they adopted to reduce discomfort now generates even greater anxiety about the prospect of working without it.

The ramifications extend throughout the academic ecosystem. Educators face unprecedented challenges in assessment validity and pedagogical design. How do you grade an essay when you cannot be certain whether the student or the AI wrote it? How do you design assignments that cannot be solved by simply prompting an LLM? These questions force educators into an arms race with technology, often resulting in assessments that measure compliance with anti-AI protocols rather than actual learning.

The Existential Threat to Mental Discipline

Though AI may function effectively as an interactive Wikipedia, the wholesale delegation of cognitive effort represents an existential threat to the cultivation of mental diligence and discipline. We are not simply changing the tools students use; we are potentially eliminating the conditions necessary for developing the capacity for deep, sustained, independent thought.

This is not a Luddite argument against technology. The question is not whether AI should be banned from education, but whether we are清醒地认识其局限性 and designing educational experiences that preserve the essential elements of intellectual development. An interactive learning companion can be valuable, but only when used to support rather than replace the productive struggle that genuine learning requires.

A Call to Intellectual Bravery

If you are a student who has dug yourself into a hole because your reliance on AI has extended beyond using it as an interactive learning companion to the realm where it handles all the heavy lifting—whether on assignments, personal projects, or other tasks—now is the time for bravery.

Stop transforming yourself into a shell of your own mind. Outsourcing everything meant to train you onto an LLM prompt is like hiring someone else to lift weights at the gym while you watch. The weights get lifted, but you get weaker. Begin the hard work of retraining your intellectual muscle, even if the initial effort feels heavy, tough, and unforgiving.

Those first weeks at the gym are genuinely difficult. Every movement reminds you of your weakness. The temptation to quit is constant. But those who persist develop real strength. Intellectual work follows the same pattern. The discomfort you feel when struggling with a difficult problem without AI assistance is not a sign that something is wrong—it is the feeling of your mind growing.

The path forward requires recognizing that AI can be a good instructor but a terrible educator. It can provide information, explain concepts, and offer guidance. But education—the development of genuine understanding, critical thinking, and intellectual resilience—requires the very struggle that AI makes unnecessary. We must choose, deliberately and consistently, to embrace that struggle, because the alternative is a future in which we have access to all the world's information but have lost the capacity to think for ourselves.