This page carries the detail behind our fourth focus area. For the argument in brief, start with Human Flourishing in AI Societies.
What Flourishing Actually Means
Serious policy and serious education both require a definition that can be argued with. Flourishing is not the same as pleasure. It is not income. It is not longevity. Each of these matters, and none is sufficient.
For the purposes of this research program, we define flourishing as a life shaped by five conditions: agency, meaning, creativity, authentic connection, and dignity. We offer these as a working account rather than a settled one, and we expect to defend them. They are not abstract virtues pinned to a wall; they are lived capacities. Each can be strengthened by design choices, and each can be weakened by the very systems that claim to optimize learning, productivity, or engagement.
Precision matters because AI does not affect all dimensions in the same way. A tool can improve short-term performance while reducing agency. It can increase output while thinning meaning. It can scale communication while degrading connection. We need language that lets us describe those tradeoffs without confusion.
Agency: The Capacity to Shape One’s Life
Agency is more than selecting from options someone else prepared. Real agency means shaping the option set, setting your own goals, and changing course when your values require it. It depends on autonomy, competence, and social recognition that your choices carry weight.
AI can weaken agency through quiet substitution. Recommendation systems narrow what people see. Predictive systems pre-sort opportunity. Interfaces learn which prompts produce compliance and repeat them. None of this requires overt coercion. A user can feel free while being steered.
Cognitive dissonance explains why this matters. A person who says “I decide for myself” but repeatedly accepts machine defaults faces tension. The fastest way to reduce tension is to accept a new identity: “I am someone who delegates judgment.” That move can happen gradually and without reflection.
The same mechanism can work in a better direction. If systems require explanation checks, invite contestation, and give users real override power, people repeatedly practice judgment. The identity shift then runs the other way: “I use tools, but I remain accountable for decisions.” That is agency preserved through design.
Meaning: The Sense That One’s Life and Activities Matter
Meaning comes from sustained purpose, contribution, growth, and relationships that acknowledge our value. A meaningful life is not always comfortable. It is one where effort connects to something that matters.
AI can undermine meaning when it automates not only tasks but status. If systems perform the work that once signaled competence and contribution, people do not just lose income. They can lose role, rhythm, and social place. The risk is a society with higher output and thinner purpose.
Again, behavior precedes belief. If a teacher is pushed to rely on automated lesson logic every day, the claim “my professional judgment matters” starts to conflict with experience. Over time, many will resolve that conflict by lowering expectations of their own role. Students do the same when every challenge is pre-solved.
The response is not blanket rejection of automation. It is institutional redesign. We need social recognition and material support for forms of human contribution that remain essential: mentoring, care, ethical deliberation, contextual judgment, and creative synthesis across differences. If these are treated as secondary, meaning erodes. If they are treated as core, automation can relieve drudgery without emptying life of purpose.
Creativity: The Generative Capacity to Imagine and Create
Creativity is not rare genius. It is everyday adaptation: a teacher changing an explanation in real time, a student reframing a hard question, a community inventing local solutions under constraint. Creativity joins novelty to context.
Generative models produce useful outputs, but usefulness is not the same as human creativity. Most generated content is pattern recombination at scale. The practical risk is cultural: we start rewarding plausible output instead of situated understanding. Quantity rises while standards blur.
Cognitive dissonance appears here too. If institutions praise originality but reward rapid reuse, people learn what is actually valued. They adapt behavior first, then revise belief: “creativity means fast assembly.” That is a loss, even when productivity metrics improve.
Education needs a different bargain with these tools. Use AI for iteration and exploration, but preserve human responsibility for problem framing, value judgment, and final accountability. Teach learners to distinguish novelty from insight, style from substance, and correlation from understanding. Creativity survives when craft remains visible and effort remains meaningful.
Connection: Authentic Bonds With Other Humans and the World
Humans become themselves in relationship. We test ideas with others, revise ourselves through encounter, and build meaning through shared effort. Connection is not just contact frequency. It requires reciprocity, risk, and the possibility of being changed.
AI mediation can help connection across distance, language, and schedule. It can also flatten it. Matching systems reduce surprise. Synthetic companions provide affirmation without mutual obligation. Engagement ranking favors intensity over understanding. People can feel constantly connected while rarely known.
The dissonance pattern is familiar. We claim to value deep relationships, then repeatedly choose low-friction substitutes. Eventually identity adapts: “this is enough.” The bar drops quietly.
Design can push in the other direction. Tools can create conditions for real dialogue, collective work, and accountable collaboration. Schools can prioritize discussion, co-creation, and conflict navigation rather than perpetual individual interaction with platforms. Technology should widen human encounter, not replace it with simulation.
Dignity: The Recognition That Persons Are Ends, Not Means
Dignity is the baseline condition for all other aims. It means persons are ends in themselves, not instruments for optimization. People do not lose dignity when they are unproductive, inconvenient, or hard to classify.
AI systems often pressure this principle through surveillance, manipulation, and reduction. Continuous tracking turns persons into behavioral streams. Targeted nudges can convert consent into choreography. Scoring systems collapse lived complexity into labels that follow people across institutions.
Here the moral issue is direct. If a system treats students as data exhaust for model improvement, it conflicts with any claim that education serves their development. If people repeatedly accept that arrangement, the dissonance is often resolved by normalizing extraction.
Dignity protection therefore cannot rely on disclosure alone. It requires constraints in architecture and policy: strict data minimization, local processing where possible, explicit purpose limits, rights to contest and delete, and governance where affected communities can refuse harmful deployments. Respect must be built into infrastructure, not added in public relations language later.
Why Education Bears Special Responsibility
Education does more than transfer information. It trains habits of thought, standards of evidence, social norms, and civic identity. In an AI-rich world, schools are not just preparing students to use tools. They are preparing students to govern tool-mediated lives.
That means education must make deliberate choices about where to use automation, where to resist it, and where to introduce productive dissonance that strengthens agency rather than conformity.
From Skill Acquisition to Capability Cultivation
Measured skills still matter, but they are no longer the full target. When systems can execute procedures quickly, schools must prioritize capabilities that remain decisively human: ethical judgment under uncertainty, cross-domain synthesis, collaborative reasoning, metacognitive control, and resilience in the face of ambiguity.
This is not nostalgia or anti-technology rhetoric. It is practical adaptation. The comparative advantage of educated humans will depend less on routine recall and more on interpretation, responsibility, and wise action in context.
A useful pedagogical strategy is guided dissonance. Ask students to commit to a position, confront them with stronger evidence, and require revision in public. That practice teaches intellectual humility and adaptive judgment. It also counters identity rigidity, which is one of the most damaging effects of polarized media ecosystems.
From Individual Competition to Collective Capability
School systems built around ranking and scarcity assume that intelligence must be sorted and rationed. AI changes that assumption. Cognitive support becomes more available, while coordination, trust, and shared judgment become scarcer.
Educational design should reflect that shift. Students need repeated practice in dialogue across difference, collaborative inquiry, and collective problem definition. Assessment should include contribution to group understanding, not only individual output.
This matters for democratic life. Citizens who cannot reason with disagreement are easy targets for identity capture. Once a person publicly fuses identity with a slogan, behavior often follows the group script. Education should train the opposite: strong commitments paired with revisability.
From Passive Consumption to Active Creation
In a high-access information environment, the bottleneck is not retrieval. It is synthesis, evaluation, and creation. Students should spend more time framing questions, testing claims, and building artifacts that show reasoning.
AI can support this shift if used as a partner in iteration. It becomes harmful when used as a shortcut around thinking. If every hard step is auto-completed, learners lose the friction that builds competence.
Dissonance is useful here as well. Students should feel the gap between “I understand” and “I can defend this claim with evidence.” Good instruction surfaces that gap and teaches how to close it through method, revision, and accountability.
From Algorithmic Efficiency to Human Relationship
Deep learning is relational. Students learn more when they are known by adults who can challenge and support them as whole persons. Families trust institutions that show care, transparency, and accountability.
AI deployments justified only by efficiency often weaken those conditions. Larger classes plus adaptive software can produce acceptable dashboards while reducing real mentorship. The alternative is clear: use AI to remove administrative drag and expand teacher time for coaching, discussion, and feedback that no system can substitute.
If schools say relationships matter but budget for automation that strips them out, students notice. They resolve the dissonance by concluding that institutions value throughput more than people. That lesson may be the most enduring one they receive.
Our Research Commitments
Our work on flourishing combines conceptual analysis, empirical inquiry, and design practice. We treat the problem as social, technical, and political at once.
Articulating Flourishing Across Cultural Contexts
We do not assume a single model of the good life. Conceptions of agency, purpose, duty, and dignity vary across cultures and institutions. We work with that variation directly through community-engaged inquiry rather than importing one normative template.
This is not a claim that all values are interchangeable. It is a commitment to pluralism with guardrails: protect dignity, preserve agency, and make tradeoffs explicit. Frameworks should be adaptable in form while firm on core protections against domination and extraction.
Measuring What Matters Without Reducing It
We develop indicators that preserve nuance: mixed methods, longitudinal observation, and participatory interpretation with educators and learners. We combine survey measures with narrative accounts, behavioral data with contextual field notes, and system metrics with lived experience.
We reject metric reductionism. What matters most is not always what is easiest to count. Some dimensions of flourishing require thick description, professional judgment, and time-based interpretation. If measurement drives the model of the person, it will eventually narrow the person to fit the model.
Designing Technology That Serves Rather Than Subverts Flourishing
We translate normative commitments into operational design principles:
Transparency: Systems should expose relevant logic and limits so users can evaluate, challenge, and override outputs.
Meaningful Control: Users should shape goals, settings, and escalation paths, not merely select from pre-scripted options.
Authentic Friction: Preserve difficulty where difficulty forms capability, including reasoning, collaboration, and creative synthesis.
Privacy as Dignity Protection: Minimize collection, localize processing when feasible, enforce purpose limits, and guarantee meaningful deletion rights.
Community Governance: Those most affected by deployment should have formal decision rights, not symbolic consultation.
Advocating for Policy That Institutionalizes Flourishing
Technology does not determine outcomes on its own. Law, procurement, labor conditions, and institutional culture decide how systems are used. We advocate for:
Right to Human Decision-Making: High-stakes school decisions, including placement, discipline, and graduation, must include accountable human review.
Limits on Surveillance: Students need protected spaces to test ideas, fail, and grow without comprehensive tracking.
Social Support Infrastructure: As automation changes labor demand, social policy should protect material security and time for non-market contribution.
Educational Mandates: Public education should cultivate broad human capability, not only outcomes legible to platform analytics.
Why This Matters: The Civilizational Stakes
The stakes here are both practical and long-range. Decisions taken this decade will shape institutions well beyond it, because defaults, once set, are rarely revisited. Four consequences seem to us especially clear:
Whether Democracy Survives: Democratic life depends on citizens who can evaluate claims, resist manipulation, and deliberate across disagreement. Populations trained into passive algorithmic dependence cannot sustain self-government.
Whether Life Retains Meaning: If institutions automate away valued roles without creating new pathways for contribution, societies may gain efficiency while losing purpose.
Whether We Remain Recognizably Human: Creativity, moral reasoning, relational depth, and self-transcendence are cultivated capacities. If we stop practicing them, they weaken.
Whether Future Generations Forgive Us: We are encoding norms now about agency, dignity, labor, and value. Future generations will inherit those defaults long after current leaders are gone.
Return to Human Flourishing in AI Societies, or see the other focus areas: Knowledge & Intelligence, Systems & Complexity, Educational Equity.