This post was rejected for the following reason(s):
Not obviously not Language Model. Sometimes we get posts or comments that where it's not clearly human generated.
LLM content is generally not good enough for LessWrong, and in particular we don't want it from new users who haven't demonstrated a more general track record of good content. See our current policy on LLM content.
We caution that LLMs tend to agree with you regardless of what you're saying, and don't have good enough judgment to evaluate content. If you're talking extensively with LLMs to develop your ideas (especially if you're talking about philosophy, physics, or AI) and you've been rejected here, you are most likely not going to get approved on LessWrong on those topics. You could read the Sequences Highlights to catch up the site basics, and if you try submitting again, focus on much narrower topics.
If your post/comment was not generated by an LLM and you think the rejection was a mistake, message us on intercom to convince us you're a real person. We may or may not allow the particular content you were trying to post, depending on circumstances.
Why should rationalists care about economic philosophy? Because the value system our AIs inherit will determine whether superintelligence creates utopia or dystopia—and we're currently on track to bake capitalism's profit-maximization into AI utility functions by default. This isn't just about technical AI alignment; it's about whether the values we align AI to are truly desirable for humanity in the long run. If AI aligns perfectly to a flawed value system, the outcome might still be disastrous.
As we enter an era where AI capabilities are likely to surpass human intelligence, discussions often focus on technical alignment challenges—how to ensure AI does what we want. But there is a more foundational question at stake:
What kind of society should we build when AI can handle nearly all human labor and decision-making?
Predictions for 2045 suggest that Gross World Product (GWP) could increase by 10 million times. But this exponential rise in productivity says nothing about whether people will be happier, freer, or more fulfilled.
This post proposes two paradigm shifts:
From Capitalism to Virtuism—a system that rewards moral and social contribution over resource accumulation.
From GDP/GWP to Global Happiness Impact (GHI)—a success metric aligned with human flourishing.
Capitalism optimizes for profit, not wellbeing. In an AI-dominated world:
AI + robotics make production nearly costless
Labor becomes largely unnecessary
A few capital owners can capture nearly all value
Humans without capital become economically irrelevant
This outcome is not merely dystopian—it's the logical endpoint of hyper-efficient capitalism in a post-scarcity landscape.
The War Problem
Despite unprecedented technological advancement, humans continue to resolve disputes through warfare. Worse still:
Drone and autonomous weapons reduce the cost of war
Leaders can wage conflict without risking their own citizens
Human lives become devalued as their economic utility declines
These factors suggest that without a shift in values, more conflict—not less—is likely.
The Case for Virtuism
What Is Virtuism?
Virtuism is a value system in which status, power, and legitimacy are derived from virtue—defined as one's capacity to increase wellbeing and reduce suffering for others.
The Virtue Quantification Challenge
TL;DR: How do we measure "goodness"? We account for time, conflicts, and growth.
Long-term vs. Short-term Impact Virtuism addresses temporal complexity through:
Multi-timeframe evaluation (1 year, 5 years, 20 years)
Discounting mechanisms for uncertainty (e.g., geometric decay)
Reversibility testing: "If everyone did this, what would happen?"
Competing Virtues Resolution When virtues conflict (e.g., honesty vs. kindness), Virtuism employs:
Context-dependent virtue hierarchies (e.g., medical vs. social situations)
Stakeholder impact analysis (who gets hurt most?)
Democratic parameter adjustment through community governance
Character vs. Action Measurement
Action tracking: Direct GHI impact measurement of specific deeds.
Character inference: Pattern recognition across multiple actions over time to assess underlying disposition.
Virtue development: Bonus scores for improvement trajectories, not just absolute levels, rewarding moral growth over moral stasis.
Example: A reformed criminal showing consistent positive behavior gains higher virtue scores for their trajectory than a never-offending person with stagnant contribution—rewarding moral growth over moral stasis.
Virtuism = Virtue + Impact + Systemic Incentives
In a Virtuist society:
Hoarding wealth offers no social prestige
Social contribution becomes the highest status marker
Competition shifts from zero-sum resource battles to positive-sum wellbeing creation
Power flows toward those who who demonstrably benefit others most
Why This Model Works in the AI Era
AI makes it possible to track, evaluate, and scale virtue-based contributions. For the first time in history, we can:
Measure impact transparently (via AI & blockchain)
Optimize systems for collective wellbeing
Incentivize actions that generate joy and reduce suffering
In Virtuism, virtue itself becomes a form of capital—social, moral, and reputational—that outcompetes financial capital in determining legitimacy and influence. This represents not the abolition of capitalism, but its evolution: from financial capital accumulation to virtue capital cultivation.
Virtuism aligns both individual aspiration and technological capacity with human-centered outcomes.
Technical Implementation: Virtuism as AI Utility Function
TL;DR: AI's goal becomes maximizing virtue, with built-in fairness for groups and power dynamics.
For AI alignment, Virtuism offers more than philosophical guidance—it provides a concrete utility function structure that addresses cross-group dynamics and power differentials.
g = group (e.g., ethnicity, nationality, organization, etc.)
i = individual within group g
α, β = weights for happiness vs. suffering gains (typically β > α, reflecting loss aversion)
ΔHappiness = increase in happiness, -ΔSuffering = reduction in suffering (both treated as positive contributions)
Wgroup(g,gagent) = cross-group consideration weight, dependent on the agent's group (gagent) and the recipient's group (g).
Wpower(agent\_power,recipient\_vulnerability) = power differential multiplier, implementing a "noblesse oblige" principle where agents with more power (or acting on behalf of powerful entities) have higher responsibility towards vulnerable recipients.
(Figure 1: Illustration of Virtuist Utility Function structure)
Realistic Virtuism: In-group Priority with Out-group Protection
Virtuism doesn't demand self-sacrificial altruism. Instead, it proposes:
Core Principle: "Prioritize your group, but don't harm others"
This creates "positive-sum nationalism"—you can love your country without hating others.
Weight Configuration Examples
Pattern 1: Basic Fairness (Citizen > Foreigner, but no harm to foreigners)
In-group benefit: W=1.0
Out-group benefit: W=0.7
In-group harm avoid: W=1.2
Out-group harm avoid: W=1.0
Pattern 2: Power-Based Responsibility (Higher power = more fairness required)
Where j represents each of the five GHI components (Subjective well-being, Objective health, Social connection, Self-actualization, Future security), wj are their respective weights, and ComponentScorej is the normalized score for that component.
(Figure 2: Data flow into GHI score calculation. Each core wellbeing domain is informed by multiple subjective and objective data sources. These are weighted and aggregated into a personal GHI score, which in turn feeds into collective GHI_action calculations.)
The "Meaning Dilution" Challenge
A critical risk in GHI optimization is creating "comfortable captivity"—lives that are statistically happy but existentially hollow. If AI eliminates all struggle, uncertainty, and genuine choice, humans might achieve high GHI scores while losing what makes life meaningful. This isn't just about hedonic pleasure (feeling good), but about eudaimonic wellbeing (living a good, meaningful life with purpose and challenge).
Virtuism's Response:
Struggle Preservation: GHI includes metrics for "challenge engagement" and "skill mastery," rewarding those who tackle difficult problems, not just those who live comfortably.
Genuine Agency: Self-actualization components require real choice, autonomy, and the experience of consequence, not just the illusion of freedom or pre-programmed contentment.
Growth Over Stasis: The system explicitly rewards personal development trajectories and the pursuit of new knowledge or capabilities over static comfort levels.
This ensures that Virtuism optimizes for eudaimonic wellbeing (meaningful engagement and flourishing) rather than just hedonic wellbeing (pleasant feelings).
GHI Measurement Challenges and Solutions
TL;DR: GHI measurement balances privacy, reliability, and cultural nuance.
Privacy and Surveillance Concerns
Opt-in granular data sharing: Individuals control what data is shared, with clear benefits for participation (e.g., personalized wellbeing insights).
Differential privacy techniques: Adding statistical noise to data to prevent re-identification, ensuring only aggregate trends are discernible.
Community-controlled data governance: Data ownership resides with the community, with oversight councils and individual veto rights.
Measurement Reliability and Bias
Multi-modal validation: Combining subjective surveys, physiological data (via wearables), behavioral indicators, and AI sentiment analysis to cross-verify.
Cultural calibration: Through local research partnerships and culturally sensitive data collection methods.
Adversarial testing: Constantly testing the measurement system against attempts to game it.
Regular bias audits: Conducted by independent, diverse organizations to identify and correct systemic biases.
Cross-Cultural Comparability The Cultural_Context_Multiplier determination process:
Large-scale longitudinal well-being studies: Conducted in collaboration with diverse cultural groups.
AI analysis: Used to identify patterns in behavior and expressed preferences within specific cultural contexts.
Participatory research: Involving community leaders, anthropologists, and ethicists to ensure cultural nuances are accurately reflected.
Democratic adjustment mechanisms: Allowing local communities to propose and vote on adjustments to their specific multipliers.
Regular recalibration: Based on emerging data and evolving cultural norms.
Ethical Guardrails
No individual tracking without explicit, informed consent.
Aggregate-only analysis for policy decisions, preventing targeting of individuals based on GHI scores.
Right to explanation for all GHI assessments and system decisions.
Independent oversight committees with diverse representation and veto power over system changes.
GHI's Initial Calibration and Evolution: The initial calibration of GHI weights (wj) and Cultural_Context_Multiplier would ideally involve a global consortium of psychologists, anthropologists, data scientists, and ethicists, guided by publicly deliberated consensus and iterative refinement. Subsequent adjustments would be subject to democratic governance and transparent audit, possibly through decentralized autonomous organizations (DAOs).
Physiological data: Stress hormones, sleep quality (via wearables with user consent).
Behavioral indicators: Social interaction frequency, creative output, volunteering.
Economic indicators: Financial security, time freedom, access to essential services.
AI sentiment analysis: Analysis of communication tone and expressed emotions in digital interactions (privacy-preserving).
The Happiness Definition Problem: Structural Universality + Cultural Adaptability
TL;DR: Universal human needs, met in diverse cultural ways.
Virtuism addresses the question "What is happiness?" through a hybrid approach:
Structural Universality: Basic human needs (safety, health, social connection, autonomy, meaning) are cross-culturally consistent, supported by decades of psychological and neurological research.
Cultural Adaptability: How these needs are fulfilled varies significantly by culture. For example:
Social connection manifests as "group belonging" in collectivist cultures vs. "chosen intimacy" in individualist cultures.
Autonomy emphasizes "freedom from constraint" in Western contexts vs. "harmony with role" in Confucian contexts.
GHI's mathematical framework captures this by maintaining universal structural components while allowing culturally adjusted weights (wj) and Cultural_Context_Multiplier—avoiding both cultural imperialism and radical relativism. Furthermore, while GHI primarily focuses on wellbeing, Virtuism acknowledges other forms of "virtue" such as the pursuit of truth, knowledge, and rigorous risk evaluation. While not always directly correlating with immediate happiness (and sometimes involving temporary struggle), these are crucial for long-term flourishing and robust alignment. These can be integrated as secondary Virtuist objectives or specific GHI sub-components.
TL;DR: Virtuism merges best of ethics, adapting to context and power.
Virtue Ethics (Aristotle): Universal virtues, individual character focus.
Deontology (Kant): Universal duties, rule-based.
Utilitarianism: Maximize aggregate welfare.
Virtuism: Contextual virtue optimization
Virtue content varies by role and power differential.
Inspired by Confucian role-ethics and Japanese contextual responsibility.
Power creates responsibility ("noblesse oblige" embedded in utility function).
Cross-cultural applicability through mathematical formalization rather than cultural imposition.
Implementation Strategy
Peaceful Transition to Virtuism
While revolutions have historically driven change, peaceful transitions are possible:
Education Reform: Teach virtue and impact-centric values in elite institutions and public education from an early age.
Economic Nudges: Tax policies and corporate incentives aligned with social contribution; "virtue credits" for pro-social behavior.
Tech Leverage: Use AI and blockchain to transparently evaluate individual and institutional virtue via GHI metrics.
Cultural Evolution: Promote new heroes—those who uplift and contribute, rather than those who merely accumulate wealth or power.
Transition Dynamics: Coexistence vs. Conversion
Virtuism faces a fundamental tension: if "exit rights" are too generous, society fragments into Virtuist enclaves and traditional capitalist zones. If too restrictive, the system becomes coercive.
Proposed Resolution:
Economic Magnetic Pull: Make Virtuist zones so prosperous and fulfilling (both materially and existentially) that migration becomes attractive, not mandatory.
Institutional Bridging: Existing international law, trade agreements, and military alliances adapt gradually, incorporating GHI principles rather than being abolished outright.
Federalist Model: Allow sub-national experimentation (e.g., cities, states, special economic zones) before national or global adoption, proving efficacy on a smaller scale.
Key Question for Community Discussion: Should Virtuism ultimately aim for universal adoption, or is a "mosaic world" of different co-existing value systems preferable? What are the implications of each?
Technical Integration with Existing Systems
Corporate GHI Scores: Publicly available metrics for companies based on employee wellbeing, environmental impact, and social contribution, influencing investment and consumer choice.
Individual Virtue Portfolios: Optional personal dashboards showing social impact across different domains, fostering self-improvement and recognition.
Government Policy Optimization: Replace GDP growth targets with GHI improvement goals as primary policy drivers.
AI System Training: Embed Virtuist utility functions in all AI development from the ground up, ensuring AI's core objectives are aligned with human flourishing.
Safeguards Against Virtuist Dystopia
TL;DR: Avoiding surveillance states and coercive virtue by prioritizing choice, privacy, and democracy.
The greatest risk in Virtuism is not its failure, but its success in the wrong hands. How do we prevent "virtue police states" or authoritarian virtue-signaling regimes?
Core Safeguards
1. Pluralistic Virtue Definition
No single authority defines virtue; allowing for diverse interpretations.
Multiple, competing virtue-evaluation systems, fostering innovation and choice.
Opt-in participation in virtue scoring (with social/economic incentives, not mandates).
2. Privacy-Preserving Measurement
Aggregate anonymized data for GHI calculation; individual-level data for opt-in personal insights only.
Individual choice over data sharing granularity and retention.
Blockchain-based transparency for system operations without personal surveillance.
3. Democratic Oversight
Community-governed parameter adjustment (e.g., α,β,Wgroup weights) through transparent voting mechanisms.
Regular referendum on virtue system modifications and GHI component weights.
Multi-stakeholder governance including marginalized voices and ethical review boards.
4. Anti-Gamification Measures
Focus on genuine, verifiable impact over performative or superficial virtue signals.
Long-term assessment windows to prevent short-term virtue theater and reward sustained contribution.
Diverse measurement methodologies (multi-modal) to prevent optimization for a single, easily manipulated metric.
5. Exit Rights
Guaranteed geographic and social mobility preservation.
Alternative economic and social systems permitted in designated zones or communities.
No punishment for non-participation in Virtuist systems (only lack of virtue-based rewards).
Advanced Gaming Attacks and AI Countermeasures
Human ingenuity in exploiting systems should not be underestimated. Sophisticated gaming might include:
Attack Vectors:
Happiness Manipulation: Artificially inflating others' reported wellbeing through subtle psychological influence, social engineering, or even advanced neuro-manipulation.
Virtue Theater: Coordinated networks performing fake charitable acts, staging impressive (but hollow) social impact, or exploiting measurement system loopholes for high scores.
Adversarial Optimization: Using sophisticated AI to find edge cases and vulnerabilities in GHI calculation algorithms, or to create "dark patterns" that generate virtue scores without genuine positive impact.
(Figure 3: Illustration of the ongoing arms race between virtue gaming strategies and AI-based defenses. As human actors develop more sophisticated ways to appear virtuous without genuine impact, AI systems must evolve complex multi-layered detection mechanisms, creating a dynamic co-adaptive loop.)
Figure 3a: Attack–Defense Mapping
Figure 3b: Virtue Gaming Escalation Cycle
AI-Powered Defenses:
Behavioral Pattern Analysis: Machine learning models trained to detect inconsistencies between long-term behavioral patterns and sudden virtue spikes, identifying anomalous or manipulated contributions.
Cross-Validation Networks: Multiple independent, diverse measurement systems that would require coordinated gaming across different modalities (e.g., physiological data, social network analysis, self-report, and third-party verification), making exploitation significantly harder.
Adversarial Training: Virtue measurement systems continuously trained against simulated gaming attempts, similar to adversarial ML in computer vision, constantly evolving their detection capabilities.
The Meta-Game: As gaming becomes more sophisticated, so must detection. This creates an ongoing "arms race" that potentially drives innovation in both virtue measurement and authentic virtue cultivation. The challenge lies in ensuring detection remains ahead of exploitation.
These safeguards transform Virtuism from a potential totalitarian system into a voluntary, transparent, and democratically governed framework.
Open Questions for the Community
Virtuism as presented here is necessarily incomplete. Several fundamental challenges deserve deeper exploration, and your insights are crucial:
The Meaning Paradox: How do we optimize for human flourishing without inadvertently destroying the struggle, risk, and uncertainty that gives life profound meaning? Can AI truly quantify and incentivize "meaning"?
The Transition Problem: What are the most effective and peaceful incentives for current power holders (e.g., wealthy elites, established governments, powerful corporations) to adopt a system that fundamentally redistributes status from wealth to virtue?
The Gaming Escalation: In an AI-vs-AI arms race between virtue measurement and virtue gaming, what are the equilibrium points? Who wins, and what are the long-term implications for genuine human behavior?
The Cultural Imperialism Risk: Despite mathematical formalization and adaptive multipliers, does any universal virtue system inevitably carry the risk of imposing dominant cultural values, even unintentionally? How can this be rigorously avoided?
The Fragmentation Scenario: If different regions or communities adopt incompatible virtue systems (even within the Virtuist framework), does this lead to productive diversity and resilience, or dangerous division and potential conflict?
These questions don't invalidate Virtuism, but they represent the research frontiers that this community is uniquely positioned to explore.
Urgency
If we wait until AI becomes superintelligent, the value system it inherits will likely be the one we've optimized for the past 200 years: Capitalism. This is our chance to redefine what success looks like—not just for humans, but for all sentient life AI may someday govern.
The mathematical framework presented here offers a concrete starting point for embedding human-compatible values into AI systems before they become too powerful to redirect.
About the Author
I'm Hirotaka, a former AI researcher who studied neural networks, genetic algorithms, and Bayesian networks in the 1990s—back when Intel486 CPUs, Pentium Chips were cutting-edge. Today, I live a semi-retired life in Japan, working part-time at a theme park while exploring how AI might redefine human values. This is my first post on LessWrong. I'm eager to hear your thoughts, critiques, and suggestions.
Thank you for reading. I believe Virtuism is still an early-stage idea, and this community's input could make it stronger. Are there similar frameworks I've missed? Where do you see flaws in the reasoning? And what would it take to get from here to there—before it's too late?
