Individual Submission Summary
Share...

Direct link:

Sustainable AI Development Pathways

Saturday, November 7, 3:30 to 5:00pm, Property: Boston Marriott Copley Place, Floor: 3rd Floor, Room: Suffolk

Abstract

AI’s socio-ecological impacts are distributed across interconnected layers of provisioning, facility operation, and computational use, yet the evidence needed to assess them remains fragmented across disciplines, system boundaries, and assessment practices. Because these impacts unfold across tightly coupled systems while governance remains divided across actors, sectors, and partial responsibilities, there is still no unifying basis for diagnosing how impacts arise, move across interconnected layers, and generate cumulative effects across the AI ecosystem or for translating that complexity into coordinated and context-sensitive guidance. This gap is especially consequential where decision-makers must balance sustainability goals against infrastructural constraints, institutional capacity, uneven technological development, and competing policy priorities.

To address this gap, this paper develops a policy-oriented framework that integrates three components: a multi-tier architecture, a socio-ecological scope structure, and a structured diagnostic schema. The multi-tier architecture differentiates the AI ecosystem into linked functional layers, while the scope structure specifies the range and mode through which impacts are manifested, induced, and extended within and across those layers. Taken together, these dimensions make it possible to trace how impacts are differentiated, amplified, and fed back across the system, while clarifying why their significance varies across functional boundaries and contextual conditions. In this way, the framework helps render cross-layer interdependencies more legible for diagnosis and comparison.
The diagnostic schema further translates this architecture into an evaluative tool that enables integrated assessment of system dynamics, contextual conditions, impact patterns, and response capacity. In doing so, it helps identify where strategic focus is needed, which trade-offs are most salient, and how more context-appropriate sustainable AI pathways can be developed under differing conditions and constraints.

By integrating fragmented evidence within a common analytical frame, the framework supports comparative analysis across diverse contexts, informs policy-level governance guidelines, and offers a flexible foundation for a wider range of stakeholders seeking to develop more grounded and context-sensitive strategies for sustainable AI governance.

Author