A new multilevel framework argues AI will only improve quality of life when deliberately designed, tested and governed around human and planetary needs; four conversion pathways and four enabling conditions identify when AI will act as a prosocial catalyst rather than an accelerant of harm.

Main Finding

The paper proposes the ProSocial AI–Quality of Life Catalyst Model (PAI‑QOL): a multilevel, outcome‑oriented framework that explains when and how AI can catalyze “pro‑people, pro‑planet, pro‑potential” improvements in multidimensional quality of life. AI is treated as a catalyst whose net effect depends on four causal pathways (agency & capability; relational & social capital; institutional & public value; regenerative & intergenerational), a lifecycle 4T discipline (Tailored, Trained, Tested, Targeted), and four enabling conditions (Double Literacy; meaningful participation; contestability; proportionality). The model advances four testable propositions and a measurement strategy that distinguishes genuine prosocial contribution from mere intention, compliance, or branding.

Key Points

• Definition: ProSocial AI is AI "deliberately designed, deployed, and governed to generate demonstrable and fairly distributed improvements in multidimensional quality of life, strengthen human and collective capabilities, and protect the ecological conditions of future well‑being."
• 4T lifecycle discipline:
  • Tailored: fit to legitimate need, context, language, risk.
  • Trained: data, objectives, and feedback aligned with social purpose.
  • Tested: beyond benchmarks → safety, subgroup performance, autonomy, capability retention, environmental load.
  • Targeted: bounded deployment with responsibility, escalation, recourse, and exit conditions.
• Four causal pathways (interacting):
• Agency & capability: AI lowers friction, supports autonomy/competence, but risks deskilling and dependency unless capability retention is designed in.
• Relational & social capital: AI can facilitate connections, translation, and coordination but can also displace human relationships or amplify fragmentation via recommender incentives.
• Institutional & public value: (summarized from framework) AI can improve responsiveness and coordination in public services but risks shifting burdens, opaque decision‑making, and accountability gaps unless governance ensures contestability and proportionality.
• Regenerative & intergenerational: AI’s ecological footprint and lifecycle impacts must be weighed against prospective environmental benefits; intergenerational capability preservation is a core evaluative dimension.
• Four enabling/boundary conditions: Double Literacy (see below), meaningful participation, contestability (ability to challenge and revise systems), and proportionality (resource use and risk matched to expected social value).
• Evaluation lens: a 4×4 matrix combining 4 experiential dimensions (aspirations, emotions, thoughts, sensations) across four levels (individual; community/institutional; national/systemic; planetary/intergenerational).
• Propositions: The paper formulates four testable claims (explicitly details Proposition 1: agency & capability; Proposition 2: relational & social capital; the remaining two map to institutional/public value and regenerative/intergenerational pathways).
• Measurement strategy: emphasizes multidimensional indicators (subjective well‑being, autonomy/agency metrics, capability retention, social capital measures, institutional outcomes, lifecycle environmental metrics, distributional analysis) and distinguishes demonstrable impacts from stated intent or branding.

Notes on “Double Literacy” - The paper identifies “Double Literacy” as an enabling condition; within the model this denotes complementary literacies required for ProSocial AI: - AI/digital literacy: ability of individuals, communities, and institutions to understand, interrogate, and use AI systems. - Ecological (systems) literacy: capacity to understand resource, lifecycle, and planetary constraints so trade‑offs are visible and accountable. - Double Literacy supports meaningful participation, contestability, and proportional governance.

Data & Methods

• Methodological approach: problem‑driven integrative conceptual synthesis (theory construction rather than statistical estimation).
• Four development stages:
• Construct clarification (defining minimal analytical properties of ProSocial AI).
• Mechanism extraction (drawing on established social and behavioral theories).
• Multilevel integration (linking individual, community, institutional, national, and planetary levels).
• Proposition and indicator development (translating the model into testable claims and observable measures).
• Sources: interdisciplinary literatures — quality of life, subjective well‑being, capability approach, self‑determination theory, social capital, public value, HCI, responsible AI, environmental sustainability, planetary health — plus emerging empirical AI studies.
• Empirical content: no new primary quantitative data; the model synthesizes existing empirical findings and theory. Generative AI was used by the author to assist drafting/language; conceptual structure and content were author‑driven.
• Outcome orientation: the model requires multidimensional, longitudinal, and distributional measurement to capture immediate, cumulative, institutional, and intergenerational effects.

Implications for AI Economics

Practical implications and research directions for economists working on AI, policy, and public investment:

1. Broaden benefit accounting beyond productivity

   • Move past sole reliance on technical performance and GDP/productivity metrics. Incorporate multidimensional quality‑of‑life indicators (subjective well‑being, autonomy, capability retention, social capital) into cost–benefit and impact assessments.
   • Include distributional impacts (who benefits/loses), because aggregate gains can mask concentrated harms.

2. Internalize externalities and lifecycle costs

   • Account for AI’s environmental footprint (energy, water, materials, e‑waste) and rebound effects in project appraisal and valuation. Lifecycle and resource proportionality should be part of feasibility and social returns computations.
   • Incorporate environmental justice: costs of infrastructure, pollution, and e‑waste are unevenly distributed geographically and socioeconomically.

3. Reframe human capital and complementarity analyses

   • Distinguish augmentation (increased human capability) from substitution (deskilling). Labor market models must consider capability retention and task redesign that preserves human judgment and learning.
   • Estimate dynamic effects: short‑term productivity gains may reduce future human capital accumulation, changing long‑run wage/productivity dynamics.

4. Account for social capital and institutional effects

   • Include metrics for bonding/bridging/linking social capital and the role of AI in changing civic participation, trust, and institutional legitimacy. These factors feed back into economic outcomes (e.g., labor market functioning, compliance costs, transaction costs).
   • Model how AI deployment shifts transaction and monitoring costs across actors (hidden burdens transferred to vulnerable parties can create inefficiencies and welfare losses).

5. Governance, market design, and regulatory economics

   • Policies that operationalize the 4Ts and enabling conditions can alter incentives for private investment. Regulatory design should require targeted testing, contestability mechanisms, and disclosure of lifecycle costs to avoid prosocial washing.
   • Consider mandates or incentives for “targeting” (deployment bounds), funding for Double Literacy initiatives, and support for participatory evaluation to reduce information asymmetries.

6. Measurement and empirical agenda

   • Develop and validate multi‑level empirical instruments: longitudinal panels that combine subjective well‑being, capability tests, social network measures, administrative outcomes, and environmental lifecycle indicators.
   • Use quasi‑experimental designs, randomized rollouts, and difference‑in‑differences to identify causal effects across the four pathways.
   • Create standardized metrics to distinguish demonstrable prosocial outcomes from declared intentions, enabling cross‑project comparisons and better cost‑benefit estimates.

7. Investment prioritization and public goods

   • Public funding and procurement can prioritize AI that demonstrably advances pro‑people, pro‑planet, and pro‑potential objectives (e.g., AI that reduces transaction costs in social services while showing no net deskilling and low environmental footprint).
   • Some pro‑planet and pro‑potential returns are public goods (e.g., resilience, intergenerational capabilities); market incentives alone may underprovide these, justifying targeted public investment.

Suggested operational steps for economists and policymakers - Require lifecycle and distributional reporting in major AI impact assessments. - Fund experiments that measure capability retention and social capital alongside productivity. - Incorporate ecological constraints into social returns calculations; develop shadow prices for environmental externalities tied to AI infrastructure. - Design procurement contracts and subsidies with 4T and enabling condition requirements to align private incentives with multidimensional social value.

Summary takeaway The PAI‑QOL model reframes AI evaluation from narrow technical or productivity metrics toward a rich, multilevel quality‑of‑life calculus that explicitly links human agency, social capital, institutional public value, and planetary limits. For AI economics this means expanding evaluative metrics, internalizing environmental and distributional externalities, rethinking human–AI complementarities, and redesigning governance and investment instruments to favor demonstrable prosocial outcomes.