The justice constraint: AGI and the Intelligence-Constrained Production Function (ICPF)
"Image synthesis assisted by Qwen Image 2.0, an AI partner within the Global Future Nexus ecosystem."
The dominant narrative of AI economics has been one of unbounded growth—more intelligence means more productivity, which means more prosperity. Yet a new framework challenges this assumption, arguing that the relationship between intelligence and economic output is not linear but constrained by justice, legitimacy, and ecological limits. The Intelligence-Constrained Production Function (ICPF) represents a fundamental rethinking of how we model the economic impact of AGI, embedding ethical considerations not as an afterthought but as a constitutive condition of sustainable growth.
The Architecture of the ICPF
The ICPF, introduced in a 2026 paper in the journal Challenges, is not a traditional production function. It models AGI as a coupled socio-technical and ecological system, linking parameters like AGI efficiency, governance quality, and externalisation intensity to justice indicators. The framework establishes three core constraints: an ecological cap on resource use (E≤10), a fairness floor (S≥0.3), and a legitimacy threshold (Λ≥0.4). These thresholds are not arbitrary; they reflect the minimum acceptable levels of distributive and procedural justice consistent with a "safe and just operating space" for humanity.
The ICPF's core insight is that AGI efficiency (A), oversight quality (H), and externalisation intensity (ξ) interact in ways that produce systemic paradoxes. Ecological load is computed as E = (A^1.2)(1 + 2ξ)(1 - 0.5H), fairness as S = H(1 - ξ)[1 + exp(-(A - 1))]^-1, and legitimacy as Λ = H. Monte Carlo simulations (5,000–10,000 runs) reveal a sobering pattern: only about 8-15% of trajectories satisfy all three justice thresholds simultaneously. This is not a calibration artifact; sensitivity analysis confirms that the paradoxes are structural properties of the model.
The Three Structural Paradoxes
The ICPF identifies three structural paradoxes that recur across parameter variations:
The Efficiency–Sustainability Paradox: Gains in AGI efficiency increase ecological load. As AI becomes more productive, it consumes more energy and resources, accelerating environmental degradation rather than alleviating it. This is the thermodynamic reality of intelligence: computation has a physical cost.
The Local Justice–Global Externality Paradox: Fairness improvements at local scales can fail to prevent burdens from being displaced elsewhere. A region that reduces its ecological footprint may simply export its pollution to another jurisdiction, leaving the global system no better off.
The Coordination–Concentration Paradox: Stronger governance oversight enhances legitimacy up to a point, beyond which centralisation erodes it. The institutions designed to guide AGI can themselves become instruments of concentrated power.
The Material Constraint
The ICPF's emphasis on ecological limits is reinforced by research on AI's physical footprint. Current AI models already have large energy and water demands; semiconductor supply chains rely on ecologically intensive mining that could pose an existential threat to the planet. "Governments are racing to develop national AI strategies," notes a BMJ editorial, "but rarely do they take the environment and sustainability into account". The lack of environmental guardrails is no less dangerous than the lack of other AI-related safeguards.
The Cognitive Mediator
A complementary framework, the Intellectually Converged Human (ICH) framework, argues that AI's productivity impact depends on human "convergence capacity"—the ability to ground AI output in embodied judgment, metacognition, and temporal integration. The ICH framework identifies four ontological constraints that no computational system can overcome: symbol grounding (AI outputs are not "about" anything until a human grounds them), embodied cognition (data is a representation of experience, never the experience itself), metacognition (the capacity to monitor one's own knowledge states), and temporal integration (the ability to consciously re-experience the past and pre-experience the future).
The implication is stark: AI cannot be an independent variable in a production function because its outputs are not yet "about" anything until a human grounds them. The production system loses its quality control mechanism without human judgment, and any model that treats AI as an independent productive factor will inevitably overestimate its contribution.
The Governance Imperative
For Global Future Nexus, the ICPF offers a governance framework that is as sophisticated as the technology it seeks to guide. The ICPF's core message—that justice-compatible trajectories are statistically rare—is a warning. Ethical and sustainable AGI outcomes do not arise spontaneously. They require deliberate design, transparent oversight, and institutional calibration to the safe and just operating space for humanity.
The path forward requires a shift from treating AGI as an exogenous productivity shock to embedding it within the ecological and social constraints that define sustainable development. As the ICPF's authors conclude, justice becomes not merely a philosophical critique but an empirical parameter—one that calibrates technological ambition to the biophysical limits and social contracts of a shared planet. The question is not whether we can build AGI, but whether we can build the governance frameworks to ensure that the intelligence we create serves human and planetary flourishing.
Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)
Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)
