``Down to Earth'': Design Considerations for AI for Sustainability from the Environmental and Climate Movement

As the Earth's temperature continues to rise increasing investments are being made to develop artificial intelligence (AI) technologies to address the current climate crisis. Through interviewing 19 participants--comprising climate and environmental advocates and developers of AI for sustainability in the US and Canada--we examine how advocates perceive and use these technologies, and how their perspectives converge and diverge from practitioners developing AI for sustainability. We identified three key findings: 1) while approaches differ, developers and advocates expressed care for people and the planet; 2) the developers and advocates values and perceptions of AI technology varied, especially around ethical issues; and 3) developers and advocates had distinct approaches to using and designing AI and digital tools. Our findings, guided by a climate justice lens, underscore the need for AI developers to: engage with advocates from intended beneficiary communities early in the design process; prioritize the urgency of the climate crisis; and emphasize tangible environmental and societal impact of digital systems.