A strong sustainability approach to AI development

2026-05-15 · Source: Nature Machine Intelligence · Field: Science & Research — Artificial Intelligence & Machine Learning, Environmental Science & Earth Systems, Research Methodology & Innovation · Depth: Expert, quick

Summary

Daniel W. O’Neill and Felix Creutzig propose a "strong sustainability" approach for assessing and guiding AI development, arguing against the limitations of traditional cost–benefit analysis. While acknowledging AI's transformative potential and existing concerns about its environmental footprint and social inequality, the authors contend that a broad cost–benefit framework, as advocated by researchers like Bossert and Loh, cannot adequately capture the complex, multidimensional nature of sustainability. Their proposed strong sustainability paradigm, rooted in sustainability science, posits that different forms of capital (natural, built, human, social, financial) possess intrinsic value and critical thresholds that must not be breached. This contrasts with weak sustainability, which permits substitution among capital forms as long as total capital value is maintained. The authors illustrate this concept with "Fig. 1: Applying the ‘Doughnut’ of social and planetary boundaries to evaluate scenarios of AI development."

Key takeaway

For AI Ethicists and Research Scientists evaluating AI's societal and environmental impacts, you should adopt a "strong sustainability" framework rather than relying solely on cost–benefit analysis. This approach recognizes that natural and social capital have intrinsic value and non-negotiable thresholds, preventing their substitution for economic gains. Your assessments must consider the multidimensional nature of sustainability to guide AI development responsibly, ensuring it operates within planetary and social boundaries.

Key insights

Strong sustainability, not cost-benefit analysis, is crucial for assessing AI's complex environmental and social impacts.

Principles

• Capital forms have intrinsic value.
• Critical thresholds must not be crossed.
• Capital cannot be reduced to monetary metrics.

Method

The proposed method involves applying a "strong sustainability" framework, exemplified by the "Doughnut" of social and planetary boundaries, to evaluate AI development scenarios, ensuring respect for environmental limits and human wellbeing.

In practice

• Evaluate AI against planetary boundaries.
• Assess AI's impact on human wellbeing.
• Avoid reducing capital to monetary value.

Topics

• AI Development
• Strong Sustainability
• Weak Sustainability
• Cost-Benefit Analysis
• Planetary Boundaries

Best for: AI Scientist, AI Ethicist, Research Scientist

Related on AIssential

• Can we AI our way to a more sustainable world? — AI is a powerful, dual-edged tool critical for solving the climate crisis, requiring careful steering and data-driven application.
• The Pope just weighed in on AI — AI's rapid advancement necessitates urgent ethical governance, robust safety measures, and clear economic justification.
• Is AI's Energy Bill Worth It? | Satya Nadella — AI's energy cost is justified only by its societal value in productivity, growth, participation, and health.
• 8 principles from human ecology can help AI work for human well-being — AI development must integrate human ecology principles, stakeholder co-creation, and ethical frameworks to ensure human and environmental well-being.
• Weekly Review 27 February 2026 — AI's pervasive impact spans ethical dilemmas, job displacement, environmental concerns, and technical advancements.
• AI for Good — The "AI for good" narrative often masks technosolutionism, weak evidence, and harmful practices by large tech corporations.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Nature Machine Intelligence.