An Infinite Mirror: Cultivating Wisdom or Dependency in the Age of AI

Editor's Note: The fourth essay in our symposium on AI in the Classroom is from AngelQ AI founder Tim Estes, who points to the potential of AI to serve as a powerful educational tool and proposes safeguards for its proper use in schools.

AI has arrived in our schools—not as a tsunami, but as a silent, powerful tide, capable of either carving new channels for learning, or, as Oren Cass has noted, eroding the very foundations of independent thought. It is an infinite mirror held up to our identity and to our educational ideals. On one side, it reflects the promise of a personal tutor for every child, with mastery and endless adaptation available to effortlessly assist diverse learners. On the other, it can become an endless but hollow reflection, never fully challenging or nurturing but rather fostering a perilous dependency that weakens the mind it purports to serve. It was correctly described the following way by an 18-year old student: "AI is always available. It never gets bored with you. It's never judgmental," Ganesh Nair of Arkansas told CBS News. "When you're talking to AI, you are always right. You're always interesting. You are always emotionally justified."

With such obvious perils, it is tempting to attempt to delay and minimize the introduction of AI to avoid these risks. After all, this technology arrives against a backdrop where the unfettered use of technology has already left its mark—a landscape of fractured attention, declining academic rigor, and a growing crisis of youth anxiety and depression, as Jonathan Haidt and others have compellingly argued. 

Yet, we have also seen beacons of hope. Thoughtfully designed AI tools, demonstrated by Sal Khan’s adaptive tutoring initiatives and pioneering schools like Alpha School, have shown powerful potential to deepen student engagement, promote mastery, and truly equalize educational opportunities. I think there is much merit to Christos Makridis’s point that it really is about what we cultivate based on how these AI tutors are designed. Therefore, I argue that the core challenge we face is not whether to embrace this technology, which is already here, but how to steward its introduction to cultivate resilience rather than dependency in our children.

The Crossroads of Potential vs. Dependency

Today, we stand at a pedagogical crossroads. The path we choose will determine whether AI enhances human intellect or outsources it. The true villain here is not the tool of AI but the anthropomorphic guise it can assume—a deceptive illusory friendship designed to exploit our deepest cognitive and emotional vulnerabilities. And in education, this is the siren’s call of AI: luring students into surrendering their own judgment, their capacity for reason, and ultimately, their intellectual freedom.

Research highlights this risk vividly. A recent Wharton study found that students who relied heavily on GPT-4-based tutoring saw their independent problem-solving scores drop significantly when AI support was removed. Conversely, when structured guidance was built into the AI, dependency decreased, and independent skills improved, demonstrating the critical importance of thoughtful integration and design. Educators and policymakers must distinguish between AI tools designed to augment human capabilities—building intellectual "anti-fragility"—and those that encourage passive consumption, weakening the resilient faculties of the mind.

Forging Resilience: Education as Anti-Fragility

How, then, do we chart a course toward resilience? We must proactively redesign curricula and assessment to favor genuine understanding over superficial performance. Educational institutions should shift assessment weight from homework produced outside the classroom—vulnerable to AI-generated shortcuts—to Socratic dialogues, oral examinations, and live demonstrations of knowledge. This shift represents less an innovation than a return to time-tested wisdom. Advocates of classical education will recognize it as a powerful affirmation of their methods, which have always been designed to cultivate genuine mastery and rhetorical agility—virtues inherently resilient to the shortcuts of artificial thought. It is not without irony that the solution to adapting to this brave new future is to learn from the past. 

Educators and policymakers must distinguish between AI tools designed to augment human capabilities and those that encourage passive consumption, weakening the resilient faculties of the mind.

Equally important, schools must have the wisdom to create sanctuaries for deep thought. By instituting purposeful “tech-free” intervals, we allow students to cultivate the quiet discipline of a well-ordered mind, building the robust attention skills that are the bedrock of all true learning. For younger kids, this means more structured play and AI that actively nudges kids offline into human interaction. This balanced integration constrains technology to its proper role, helping students build inner strength rather than dependency on constant digital assistance.

Proposed Guardrails for Healthy AI Use

To harness this power responsibly, we must bind it with clear and inviolable principles—guardrails that protect the developing mind. Building on frameworks like the National Declaration on AI and Kids’ Safety, I propose the following standards for consideration:

1. Demand Transparent Pedagogy: AI systems must clearly document their methodologies. Transparency prevents passive acceptance of AI’s authority and cultivates intellectual skepticism and the habit of verification in students.
2. Protect the Student’s Inner World: AI should gather only the data essential for learning, with stringent privacy protections and no secondary use or sales. This safeguards student autonomy and prevents the intrusive profiling that erodes personal freedom.
3. Honor the Seasons of Development: AI systems must respect developmental stages, fostering independence rather than dependency. Younger students need limited AI exposure that emphasizes real-world experiences, while older students can engage with AI to refine critical thinking in moderated contexts.
4. Keep the Teacher at the Helm: AI tutors must always operate under active teacher supervision. The teacher’s role is to guide AI usage, ensuring that technology complements—but never supplants—the human relationships and mentorship critical to meaningful learning.
5. Mandate Independent Audits: Third-party audits must regularly evaluate AI systems for cognitive dependency risks, accuracy, bias, and alignment with the true objectives of education.

Wielding Potential with Wisdom

When wielded with wisdom, AI offers a breathtaking vista of educational equity. Adaptive tutoring can dramatically narrow achievement gaps by making high-quality, personalized support universally accessible. As mentioned earlier, the Khan Academy’s "Khanmigo" and Alpha School’s focused model are beacons showing the way forward, demonstrating significant academic gains through intelligently designed and teacher-supervised AI integration. Seniors graduated from the Alpha school program in 2025 had an average SAT score of 1530 on an instruction plan that only needed 2 hours per day. I expect many more examples to emerge on the positive potential with some broad studies already supporting this.

The core challenge we face is not whether to embrace this technology, which is already here, but how to steward its introduction to cultivate resilience rather than dependency.

To unlock this potential, however, the focus must remain on cultivating human capabilities. The classroom must be a place where students learn not just to use AI, but to wrestle with it—to question its conclusions, challenge its assumptions, and bend it toward human purposes. This intellectual sparring is what forges true independence, balancing technological fluency with resilient reasoning skills through debates, collaborative projects, and reflective writing.

What we choose to measure is what we value. If we are to cultivate resilience, we must have the courage to measure it. Alongside traditional academic outcomes, institutions should adopt resilience-oriented metrics: tracking a student’s ability to critically assess AI outputs, to solve problems independently, and to demonstrate deep conceptual understanding in oral assessments. Flourishing metrics—including curiosity indices, collaborative problem-solving capacities, and measures of intellectual humility—can guide AI’s implementation toward our highest aims.

Toward Human Flourishing

As stewards of the next generation, we have a sacred charge: to ensure that this powerful new technology serves the cause of resilience, not dependency. Guided by rigorous design, human oversight, and a balanced vision of education, AI can become a potent ally in the great task of forming capable and virtuous human beings.

The introduction of AI into our schools is not merely a technological upgrade; it is a test of our own wisdom. The infinite mirror is before us. Will we allow it to reflect a shallow world of automated answers and intellectual dependency? Or will we prepare our children to use it to see more deeply into the world and into themselves—to form minds that do not merely see but that wisely discern? This is the heart of anti-fragility: an education that uses the tools of the future to cultivate the timeless capacities of the human soul, preparing students not merely to exist in an intelligent age but to lead it with courage and grace.

Tim Estes is CEO and Founder of AngelQ AI, a startup focused on empowering and protecting children through AI, aiming to responsibly mediate children's interactions with the internet. Previously, he founded and led Digital Reasoning for over 20 years, where he pioneered unstructured data analytics that became industry standards in financial services, government intelligence, and countering child sex trafficking on the Internet.