Artificial intelligence has favourable potential in crucial fight against global warming. However, the challenge is limiting AI’s own contribution of greenhouse emissions in its
development and utilization.
Discussions at COP 26 and recent IPCC reports on climate emergency
call for large multistakeholder mobilization including actors across politics, business,
science and civil society. As a direct result, the GPAI Working Group on Responsible Artificial Intelligence has created a Committee on Climate Action and Biodiversity Preservation, Co-Chaired by Nicolas Miailhe of The Future Society and Raja Chatila of Sorbonne University, to start addressing questions regarding AI’s potential to support climate action.
Funding multidisciplinary innovation
The Working Group has partnered with Climate Change AI (CCAI), a global initiative that brings climate experts together with their counterparts in AI, and with the UK-based Centre for AI and Climate (CAIC) to develop its initial climate roadmap in the responsible adoption of AI for climate action.
Artificial intelligence is already helping to anticipate future needs for solar energy generation, optimise the construction of energy-efficient buildings, identify areas of deforestation from satellite images, and analyse company data based on its environmental benefits. Entering into the next phase of its work, the Working Group proposes that governments:
– improve access to data and digital infrastructures with the potential to apply AI principles to environmental issues
– prioritise funding for multidisciplinary research and innovation
– facilitate the rollout and integration of environmentally beneficial AI systems for energy, transport, agriculture and heavy industry
Assessing the carbon footprint of AI
Given that the overall aim is to slow the progress of climate change, the Working Group raised the concern that using AI applications can affect and even increase greenhouse gas emissions. Additionaly, AI applications have been implicated in the carbon footprint and life cycle of the hardware and software used to run them. Before its by product emissions can be reduced, the environmental impact of using AI must be assessed.
In order to complete an accurate assessment, the Working Group proposes to collect appropriate data and measure the environmental impact of AI, addition to identifying and specifying standards, assessment tools and the most virtuous practices. Given the accelerating pace of AI usage, the experts advise that governments play a proactive role in bringing together multiple entities, including digitalisation and climate change agencies, regulators, academics and the private sector.