IGF 2023 Open Forum #37 Planetary Limits of AI: Governance for Just Digitalisation?

The digital transformation increasingly contributes to greenhouse gas (GHG) emissions. For example, generative artificial intelligence (AI) applications consume energy at different stages of their lifecycle: during 1) hardware manufacturing, 2) model training, and 3) model deployment. The companies leading in the development of generative AI are reluctant to provide exact figures on energy consumption. However, first studies point to a high overall impact on the environment:  a conversation with ChatGPT consumes around half a liter of water for 25 to 50 individual questions (for cooling the data centers). The integration of AI chatbots into search engines could increase the required computing power by a factor of five – and thus also their energy consumption. As such, the increased application of AI could put an additional strain on efforts to stay within agreed boundaries and to limit climate change.

Different actors around the globe want to address risks and challenges related to AI. However, regulatory approaches focus primarily on ensuring safe use and on addressing societal challenges, such as deep fakes and disinformation. To ensure sustainable AI, there’s a need to limit the ecological and social risks. In particular, countries of the Global South need to be involved in this process, as AI training often takes place in emerging markets. Further, many countries in the Global South are rapidly developing their data economies. Here, the provision of sustainable digital infrastructure, such as green data centres, and frameworks for circular ICT can help to ensure a Just Transition. How can we ensure that sustainability aspects are considered in the development and deployment of the digital technologies, such as AI, and form the basis of the digital transformation?

The IGF forms the ideal forum to inspire policymaking by discussion and sharing good practices. In this Open Forum, stakeholders such as ITU, government representatives, civil society representatives and business will jointly discuss options for action for the sustainable development of the ICT sector and technologies, such as AI. The session brings together representatives from politics, business and civil society who work on sustainable ICT initiatives. For instance, international organisations such as the ITU and the World Bank are developing guidelines for sustainable data centers on behalf of the German Federal Ministry for Economic Cooperation and Development (BMZ). As data centers form the basis for computation and data storage that supports AI deployment, limiting their environmental impact and energy use is a first important step for AI to remain within planetary boundaries.