We could be building a better world with our students. Why are we disempowering them instead?
Source: Mike Caulfield Substack
Author: Mike Caulfield
Published: 2026-06-17
Source type: essay
Original source: https://mikecaulfield.substack.com/p/we-could-be-building-a-better-world
Private backup: the full article text is archived in the private repository at archives/articles/mikecaulfield-substack-com-we-could-be-building-a-better-world.source.md. It is not published on the public Quartz site.
Summary
Mike Caulfield argues that education has abandoned too much of its constructionist impulse just as agentive AI makes authentic, discipline-shaped student projects newly feasible. He contrasts older forms of digital constructionism—community mapping, public publishing, and real-world student work—with a present in which edtech is too often reduced to delivery, surveillance, and anxiety about AI’s harms.
His central claim is not that schools should ignore those harms, but that educators are missing a larger opportunity. Classifier projects, structured data work, and other AI-assisted builds can let students make meaningful things, audit outputs, revise definitions, and practice the kinds of judgment they will need in AI-shaped workplaces. For Caulfield, the deeper ethical question is whether education will merely warn students about AI or actually prepare them to use it in responsible, world-facing work.
Pull quotes
The missed opportunity
“The bigger issue is I think the opportunity we are missing.”
AI and constructionism
“The good thing is that AI, especially agentive AI, is incredibly well-suited to constructionism projects.”
Iteration as learning
“Build the classifier, audit the output, rethink and rewrite the classifier.”
Educational responsibility
“It seems to me that given the debt we put these students in… that we might talk less about AI ethics, and more about the ethics of our educational responsibilities.”
Durability note
The specific examples—film classifiers, newspaper descriptors, and weather-event projects—may date as tools change, but the broader argument is durable: AI can support authentic, inspectable student work when the learning design keeps purpose, debate, and revision in students’ hands.
Big ideas
- AI literacy has to be taught inside real subjects
- AI is changing what knowledge work asks people to do
- Students need to bring the purpose; AI should not supply it for them
Claims
- AI literacy should teach students what to do with AI, not just what to think about it
- AI literacy only works when it is connected to subject-area knowledge
- Agentic AI can preserve thinking when students have to design the work
- AI-built models can help students show conceptual understanding
Key evidence and examples
- Caulfield revisits constructionism as learning-through-making, using examples such as archive projects, lake-cleanup chemistry, board-game economics, and community resource mapping.
- He argues that older digital constructionism lost momentum as online self-publication became riskier, students became more reluctant to publish publicly, and COVID-era edtech hardened into a delivery model.
- He proposes agentive AI as a better fit for renewed constructionist work because students can build classifiers, inspect legible data structures, and iterate on definitions without needing deep programming expertise.
- His examples include classifying film genres, tracking gendered modifiers in historical newspapers, and pulling and categorizing extreme-weather events over time.
- In each case, the learning value comes from debating categories, auditing outputs, revising prompts/classifiers, and explaining how disciplinary judgment shaped the final result.
- He argues that this kind of work better prepares students for AI-shaped workplaces than simply amplifying fear, prohibition, or abstract ethics talk.
Education relevance
This is highly relevant to higher education, AI literacy, disciplinary pedagogy, project-based learning, authentic assessment, and workforce preparation. It offers a concrete bridge between classroom AI use and the kinds of supervised, iterative, judgment-heavy work students may encounter after graduation.