Divya van Mahajan

Will AI Destroy or Reinvent Education?

ai-daily-brief-podcast

Will AI Destroy or Reinvent Education?

Overview

This episode of The AI Daily Brief (published August 17, 2025) is a “long read / big think” episode in which the host synthesises three published essays on AI’s impact on education, then advances his own argument. The central thesis is that the debate over AI in education cannot be resolved without first settling a deeper, older question: what is the purpose of education? The host argues that education has always served two competing goals — teaching people how to think and how to be in the world, versus teaching economically productive skills — and that AI is accelerating a crisis that existed long before ChatGPT, while also forcing an overdue reckoning.

Source video URL was not provided.

Prerequisites

  • Basic familiarity with large language models (LLMs) and tools such as ChatGPT
  • Awareness of the student loan/debt crisis in the United States
  • General understanding of liberal arts versus vocational education traditions
  • Familiarity with concepts such as cognitive load, neuroplasticity, and EEG measurement is helpful but not required
  • Some context around the history of coding boot camps and alternative credentialing

Main Points

1. The MIT Study: AI Externalises Cognition

  • A small, non-peer-reviewed MIT study recruited 54 participants to write essays under three conditions: AI-assisted, search-engine-assisted, or brain-only.
  • AI-assisted essays contained more specific factual references but were more homogenous; brain-only essays produced greater variety of arguments.
  • Participants who used AI showed poor recall of their own writing — they had not internalised it.
  • EEG measurements showed brain-only writers had the highest neural connectivity across brain regions; AI users showed up to 55% lower connectivity by the DDTF (dynamic directed transfer function) measure.
  • Researchers concluded that external support tools “restructure not only task performance, but also the underlying cognitive architecture.”

2. David Brooks — “Are We Really Willing to Become Dumber?”

  • Brooks positions himself as an AI optimist but warns of a “malevolent seduction”: the illusion of excellence without effort.
  • He frames AI-assisted thinking as “empty calories for the mind” — it produces output without building the cognitive muscle that hard thinking develops.
  • His core claim is simple and intuitive: thinking hard strengthens mental capacity; outsourcing thought to a bot diminishes intellectual potential.

3. Megan O’Rourke — The Educator’s Personal Reckoning

  • O’Rourke (Executive Editor, Yale Review; creative writing professor, Yale) initially found ChatGPT genuinely liberating for administrative and logistical tasks, describing it as having “the cheerful affect of a golden retriever and the speed of the Flash.”
  • The pivot: over time, the tool’s availability began to interfere with her own thinking — AI-drafted versions of documents she could have written herself “lodged unhelpfully” in her mind.
  • She sent an email drafted by AI that she later realised did not express what she actually felt — describing the experience as “a ghost with silky syntax colonising my brain.”
  • Her deeper concern is not skill loss but the loss of a mode of being: the pleasure of invention, the “felt life of the mind at work,” and the reward of pressing toward an elusive thought.
  • She proposes structural reforms: pass/fail writing classes, in-class writing labs without AI access, replacing take-home essays as assessment tools.

4. John Craycraft — The Student’s Ground-Level View

  • Craycraft (third-year student, University of Minnesota) observes a campus-wide shift from curiosity-driven learning to completion-oriented efficiency.
  • The dominant student question has become “How can I get this done fastest?” rather than “What can I learn from this?”
  • Common peer attitudes: “Why do I need to learn this when AI can do it for me?” and “I don’t need to understand the process if AI gets the right answer.”
  • His argument: college should be about developing the capacity to think, question, and navigate complexity — not accumulating facts or completing tasks.

5. The Host’s Framework: Two Competing Purposes of Education

  • Education has always served two distinct and often conflicting goals:
    1. Learning how to think — how to construct meaning from experience, exert control over attention, and be in the world (the liberal arts tradition).
    2. Learning how to do economically productive things — acquiring marketable skills commensurate with the cost and time invested.
  • This tension predates AI and is embedded in cultural aphorisms (Oscar Wilde, Mark Twain) and in structural failures like the student debt crisis.
  • Coding boot camps are cited as a historical example of the market attempting to resolve this tension outside traditional institutions.
  • The host draws on David Foster Wallace’s 2005 Kenyon College commencement address (“This Is Water”) to articulate the deeper liberal arts case: education is not about filling students with knowledge but about learning to choose what to think about and how to construct meaning from experience.

6. The Host’s Argument: AI Is an Accelerant, Not the Root Cause

  • The student loan crisis and the mismatch between education and the job market predate ChatGPT (November 2022); AI did not create these problems.
  • AI may be harmful to the first goal of education (learning to think) — the neurological evidence is worth taking seriously.
  • AI is clearly valuable for the second goal (economically productive skills) and students asking “why learn this when AI can do it?” have a legitimate claim within that frame.
  • The productive path forward is to design explicitly for both goals, identifying which parts of the educational experience serve which purpose, rather than treating them as interchangeable.
  • The host expresses optimism that AI’s disruptive force will compel a structural reinvention of education that should have happened decades ago.

Key Concepts

  • Cognitive architecture: The underlying neural structures and connectivity patterns that support thinking and learning, which the MIT study suggests can be restructured by reliance on external tools.
  • Dynamic Directed Transfer Function (DDTF): A neuroimaging analysis method used in the MIT study to measure directional information flow between brain regions; AI users showed up to 55% lower connectivity by this measure.
  • Empty calories for the mind: David Brooks’s metaphor for AI-assisted thinking — it produces output without the cognitive effort that builds genuine intellectual capacity.
  • Liberal arts education (liberal arts tradition): An educational philosophy centred on developing the capacity to think critically and construct meaning, rather than acquiring specific vocational skills.
  • Vocational/practical education: An educational model focused on teaching economically productive, job-market-relevant skills (e.g., coding boot camps).
  • “This Is Water” (David Foster Wallace, 2005): A Kenyon College commencement address arguing that the real value of a liberal arts education is learning to exercise conscious control over what one thinks about and how one constructs meaning from experience.
  • LLM (Large Language Model): The class of AI system (e.g., ChatGPT) central to this discussion, capable of generating text that simulates human reasoning and writing.
  • Homogeneity of AI-generated writing: The observation from the MIT study that essays written with AI assistance were more similar to one another, lacking the argumentative diversity of unaided writing.

Summary

The host argues that the question “will AI destroy or reinvent education?” cannot be answered without confronting a far older and unresolved tension: education has always been pulled between teaching people how to think and teaching them how to earn a living, and society has never satisfactorily reconciled these two goals. Drawing on a neurological study from MIT and essays by David Brooks, Yale professor Megan O’Rourke, and a University of Minnesota undergraduate, he presents credible evidence that AI externalises cognition in ways that may genuinely impair the development of deep thinking — and takes that evidence seriously. At the same time, he insists that students who ask “why learn this when AI can do it for me?” are raising a legitimate question within the vocational frame of education, and that empowering people to use AI as a tool for creative and economic productivity is not only acceptable but necessary. His conclusion is that the failure of education is structural and predates AI; that AI is an accelerant of both the crisis and the opportunity; and that the only productive response is to redesign educational institutions explicitly around both purposes — protecting spaces for unmediated thinking and cognition while aggressively embracing AI for skill-building and creation. He is cautiously optimistic that AI’s sheer disruptive force will compel the systemic reinvention of education that should have occurred decades ago.