Principal Investigator : Dr Proscovia Svard and Dr Yann Rodier
The pressing need for sustainability has catalyzed collaborative initiatives between public institutions and private enterprises to confront multifaceted environmental, social, and economic challenges. This research explores the integration of Large Language Models (LLMs) within governmental institutions, emphasizing their potential to enhance public service delivery while addressing sustainability concerns. The Brundtland Commission's definition of sustainable development serves as a foundational principle, advocating for practices that meet present needs without jeopardizing future generations. Despite the adoption of the United Nations' Agenda 2030 and its 17 Sustainable Development Goals (SDGs), significant hurdles remain, including resource depletion and rising inequalities, exacerbated by the rapid evolution of technologies like LLMs. While LLMs can streamline operations, improve efficiency, and enhance citizen engagement, they also pose ethical dilemmas related to data privacy, potential biases, and transparency challenges. This research aims to employ the Life Cycle Sustainability Assessment (LCSA) framework to evaluate the environmental, economic, and social implications of LLM deployment in the public sector and will utilize a mixed-methods approach, combining quantitative assessments of environmental and economic impacts with qualitative analyses of social implications. It will be structured into four work packages: (1) a literature review to gather existing knowledge on Environmental Life Cycle Assessment (E-LCA), Social Life Cycle Assessment (S-LCA), and Life Cycle Costing (LCC); (2) the development of a conceptual LCSA model; (3) case study analyses of LLM applications in various public sectors and their impact on the management of public information; and (4) the specification of an AI eco-design tool to facilitate sustainability assessments. It aims to identify gaps in current methodologies and emphasizes the importance of interdisciplinary integration to address the sustainability challenges posed by AI technologies if the public sector values are to be maintained.