[month] [year]

Data Centres for AI

The Human Sciences Research Centre in collaboration with the Global Digital Cultures Research Group, King’s College London, UK  hosted an international online workshop on Data Centres for AI: Global Expansion and the Environment on 9 and 23 October. Over 300 participants registered from 26 countries across six continents, representing a range of backgrounds including students, faculty, research fellows, software engineers, journalists, policy professionals, AI/ML startup founders, tech policy experts, social scientists, and data scientists.

 The first session hosted speakers and participants from across the globe where the theme of environmental impact of AI and Data Centres was deeply explored. The speakers included Pradip Thomas, The University of Queensland, Australia; Liza Cirolia, African Centre for Cities, South Africa; Nicolas Diaz, Future of Artificial Intelligence Research (FAIR), Chile and  Paola Ricaurte, ecnológico  de Monterrey, Mexico as the discussant.  Opening remarks were given by workshop coordinators – Dr Aakansha Natani and Sebatian Leheude, Lecturer in Ethics, AI and Society, King’s College London,UK. 

Dr. Aakansha Natani, the workshop coordinator and a faculty member at IIITH, opened the first session on 9 October by emphasizing the urgent need to balance the demand for digital infrastructure with environmental sustainability. She stressed that sustainable digital infrastructure is not solely a technical or economic issue but also a social and ethical imperative. Co-coordinator Sebastian Leheude, Lecturer at King’s College London, UK, set the tone by highlighting global trends and the increasing environmental footprint associated with data centre expansion.

Pradip Thomas from the University of Queensland discussed the emerging patterns of “greenwashing” that often obscure the environmental impact and high resource consumption of data centres. 

Liza Cirolia from the African Centre for Cities, South Africa, offered insights into data centres in the African context. She discussed Africa’s vital role in supplying minerals essential for digital infrastructure and raised questions about the trade-offs made to meet data centre demands, exploring the political and ethical dimensions of these choices. 

Nicolas Diaz from the Future of Artificial Intelligence Research (FAIR), Chile, addressed on-site challenges faced by architects during data centre construction and the thoroughness of environmental norm compliance. 

Paola Ricaurte from Tecnológico de Monterrey, Mexico, served as the discussant for the first session, covering broad themes such as state discourse on green energy, environmental degradation, and the impact on indigenous communities, with a focus on policies in India and Mexico. 

The session concluded with calls for more in-depth studies on the operation of data centres, their energy consumption patterns, and the initiation of political discourse surrounding these issues.

The second session, held on 23 October, began with Julia Velkova from Linköping University, who discussed power dynamics in Sweden, where major tech companies are pre-purchasing grid capacity, driving up costs and limiting access to electricity for local users. 

Patrick Brodie from University College Dublin examined recent developments in Ireland through the concept of “eco-modernity,” highlighting that Big Tech’s involvement in energy transition has implications for global energy governance. He distinguished between eco-socialist and eco-capitalist transitions, each with significant implications for energy policy.

 Shaolei Ren from the University of California, Riverside, advanced the argument for the social and environmental responsibility of AI. He detailed the substantial water requirements for cooling data centres and regional imbalances in water consumption. 

Sebastian Leheude, the discussant in the second session, emphasised the need for deeper research into the intersections of AI, data centres, and their environmental impacts.

The workshop concluded with recommendations for further studies on data centre operations, energy consumption, and the development of broader political discourse on these critical issues.

 

October 2024