Blueprints for AI Equity and Transparency in Climate Science

AI is now being used to augment climate simulations and improve projections regarding impacts of climate change. Parallel to the scientific breakthroughs ushered in by AI, there is a noticeable desire to augment the role of climate activists and affected communities in decision-making. As a result, an interdisciplinary body of scholarship and activism has pushed interdisciplinary research on fair, transparent, and accountable AI forward, ushering in new AI regulations and policies that mandate impact assessments, audits, AI transparency requirements, as well as community engagement.

These developments beg the question of how we can leverage the momentum gained in AI policy and in climate science to ensure equitable and sustainable AI climate innovations that serve the public interest. This is particularly pressing given that there is a discernible compartmentalization in policy realms. While AI regulation champions transparency, its focus leans more towards consumer AI rather than scientific applications.

To realize the full potential of AI in combating climate change and ensuring inclusive and equitable access to innovation, especially among those most vulnerable to climate emergency events, and to avoid exacerbating already existing social and climate disparities, two issues must be addressed:

1)  it is paramount to move from the technical aspects of AI to integrating the social dimensions of climate change, particularly the real-world experiences and expertise of affected communities into the realm of climate science and policy.

2) it is key to build on community engagement to increase meaningful AI transparency in climate science AI.

The objective of this project is to develop equitable and inclusive engagement on AI transparency in climate science and use the workshops to co-create relevant research questions to explore in future research that will support the formulation and implementation of equitable AI policies that build on local knowledge, community preferences, and explainable scientific methods. The main output of this workshop will be a blueprint for creating pathways for community-based AI transparency in climate science.