The Question Technology Forces Us to Answer
AI systems now shape daily behavior at scale. They set defaults, rank options, and steer attention. And once behavior changes, identity and belief tend to follow.
That is cognitive dissonance working in the open. If you can get someone to act against their stated values, discomfort appears—and most people resolve it not by saying “I was wrong” but by repeating the new behavior until it feels normal. In persuasion, that mechanism is powerful. In politics, it is visible at both extremes.
AI brings the same mechanism into ordinary institutions, schools among them. Students use tools that reward speed over reflection, engagement over depth, compliance over judgment. Small interaction patterns become habits. Habits become norms. Norms become identity. A student who once said I think for myself can, over time, become someone who says the system already knows what I need.
This is why AI is not only an economic or technical shift. It changes what people practice every day, and practice is what forms character. When machines take over work long treated as the mark of human distinctiveness—analysis, composition, tutoring, forms of care—one question becomes unavoidable:
What should humans keep doing, because those activities are part of a life worth living?
For the full interactive knowledge graph experience, please visit this page on a desktop computer.
Five Conditions
Serious policy needs a definition that can be argued with. Flourishing is not pleasure. It is not income. It is not longevity. Each matters; none is sufficient.
We define flourishing as a life shaped by five conditions. We offer these as a working account rather than a settled one, and we expect to defend them. Each can be strengthened by design. Each can be quietly weakened by systems that claim to optimize learning, productivity, or engagement.
Agency
The capacity to shape your life, not select from a menu someone else prepared
Eroded when recommendation systems narrow what you see and interfaces learn which prompts produce compliance. No coercion is required; a person can feel free while being steered.
Protected when systems require explanation, invite contest, and grant real override power—so judgment gets practiced rather than delegated.
Meaning
The sense that your life and your work matter to someone
Eroded when automation takes not just tasks but status. If a system performs the work that signalled competence, people lose role, rhythm, and social place—not only income.
Protected when mentoring, care, ethical deliberation, and contextual judgment are treated as core contributions rather than soft extras.
Creativity
Everyday adaptation under constraint, not rare genius
Eroded when institutions praise originality but reward rapid reuse. People learn what is actually valued, adapt, and then revise the belief: creativity means fast assembly.
Protected when humans keep responsibility for framing the problem and making the final call, and craft stays visible.
Connection
Bonds with reciprocity, risk, and the possibility of being changed
Eroded when matching systems reduce surprise and synthetic companions offer affirmation without obligation. People can feel constantly connected and rarely known.
Protected when schools prioritize discussion, co-creation, and conflict navigation over perpetual solo interaction with a platform.
Dignity
The baseline: persons are ends in themselves, never instruments for optimization
Eroded by surveillance, manipulation, and reduction—continuous tracking that turns persons into behavioral streams, scores that collapse a life into a label and follow it between institutions.
Protected by architecture rather than disclosure: data minimization, local processing, explicit purpose limits, rights to contest and delete, and communities that can refuse a deployment outright.
Why Education Bears Special Responsibility
Schools do more than transfer information. They train habits of thought, standards of evidence, social norms, and civic identity. In an AI-rich world they are not preparing students to use tools. They are preparing students to govern tool-mediated lives.
That means choosing deliberately where to automate, where to refuse, and where to introduce the productive difficulty that builds capability. When systems execute procedures instantly, the target shifts toward what remains decisively human: ethical judgment under uncertainty, synthesis across domains, collaborative reasoning, and resilience in ambiguity.
One technique we keep returning to is guided dissonance. Ask students to commit to a position, confront them with stronger evidence, and require public revision. It teaches intellectual humility and adaptive judgment—and it works directly against the identity rigidity that polarized media reliably produces.
How We Work
We do not assume a single model of the good life. Conceptions of agency, purpose, and dignity vary across cultures, and we work with that variation through community-engaged inquiry rather than importing one template. This is not a claim that all values are interchangeable. It is pluralism with guardrails: protect dignity, preserve agency, make trade-offs explicit.
We also reject metric reductionism. What matters most is not always what is easiest to count, and if measurement drives the model of the person, it will eventually narrow the person to fit the model. So we combine survey measures with narrative accounts, behavioral data with field notes, system metrics with lived experience.
And we translate all of it into design principles we can be held to: transparency, meaningful control, authentic friction where difficulty forms capability, privacy as dignity protection, and formal decision rights for the communities a deployment affects.
The Path Forward
Here is one test any school can run before signing a contract: audit the dissonance loops. Ask where a platform invites students to claim autonomy while quietly reducing real choice. Ask where teachers are told to exercise professional judgment while the dashboard penalizes deviation from it. Ask where families are promised transparency while the contract restricts independent review.
We do not read these contradictions as communication failures to be smoothed over at rollout. They are design features, and what they train is submission. Caught early, they can be fixed—workflows redesigned, contracts renegotiated, safeguards made non-negotiable. Left in place, the drift is predictable: students internalize dependence, educators narrow their ambition, and institutions come to mistake compliance for learning.
We are at a threshold, and we do not think that is overstatement. Choices about AI in education will shape not only what students learn but who students become, and what kinds of communities they are then able to build. Choices of that consequence deserve full public scrutiny and full ethical seriousness. Our contribution is to make sure they receive both.
The research behind this The five conditions in full, and the policy we argue forEach condition examined at length, the four shifts education must make, our research commitments, and the legislative and procurement changes that would institutionalize flourishing.