Artificial intelligence (AI) both reflects and amplifies the systems of power and privilege that birthed it, creating a fast-widening gap between those who design it and those who are subject to it. One of the places that gap looms largest globally is in environmental impact, where centuries of racism, colonialism, and extractivism continue to expand it.

The movement for Environmental Justice (EJ) was actually developed as a direct response to environmental racism, which is the disproportionate impact of environmental hazards on people of color, as outlined by the The First People of Color Environmental Leadership Summit. EJ is concerned with equitable distribution of environmental goods between nation-states and the impacts of uneven development around the world.

This isn’t a concern we can afford to address in some far off future; Big Tech’s AI development is already exponentially increasing energy usage, as cloud computing leaves a massive carbon footprint. The International Energy Agency estimates that data centers that power AI will double their energy use between now and 2026, putting them on par with the entire nation of Japan. While orgs in the Global North dominate the use of resources to train models — we know that training just one emits as much carbon as five cars over their entire useful lives — communities in the Global South bear the brunt of the fallout.

Then there’s the mining for conflict materials used to create semiconductor chips to power AI, defined by the U.S. Securities and Exchange Commission as the columbite-tantalite (coltan), cassiterite, gold, wolframite, and their derivatives (tantalum, tin, and tungsten) that finance war in the Democratic Republic of the Congo and adjoining countries. When we add in AI’s complicity in the spreading of climate disinformation and planned obsolescence of the hardware used to run models and individual apps clogging landfills with e-waste, we have a perfect storm for environmental damage that disproportionately impacts people who are already under-resourced around the world — and underrepresented in the AI space.

But the outlook isn’t all bleak. As a parent who is committed to leaving the world better than we found it for all of our kids, and an advocate who is committed to enabling and promoting transparent, responsible AI, I know that — when used with the well-being of the people it impacts in mind — it could potentially help us tackle environmental issues. Mozilla is committed to the battle; our 2024 cohort of the Mozilla Technology Fund centers the role open-source AI systems can play when it comes to Indigenous justice, climate change, and food and energy justice. And we’re founding members of the Green Screen Coalition to further invest in projects and people working at this intersection.

I connected with Dr. Priya L. Donti (she/they), co-founder and chair of Climate Change AI (CCAI), a global nonprofit working at the intersection of climate change and machine learning, to discuss how we can further leverage AI in this space. Dr. Donti is also an assistant professor and the Silverman (1968) Family Career Development Professor at MIT’s Electrical Engineering & Computer Science Department and Laboratory for Information and Decision Systems. Her research centers the use of machine learning to forecast, optimize, and control power grids that make use of renewable energy sources.

Rankin: Why is it important to you to conduct research at the intersection of climate change and machine learning?

Donti: Climate change is one of the most pressing issues we face today, requiring us to mobilize all the tools we have. Machine learning is one such tool. It has the potential to significantly accelerate climate action when applied in a responsible and impactful manner. For instance, machine learning has been used for applications such as forecasting renewable energy production in power grids, optimizing heating and cooling systems in buildings, creating real-time flood reports from satellite imagery to aid in disaster response, analyzing corporate financial disclosures for climate-relevant information, and accelerating the design of next-generation batteries. My own research focuses on developing robust machine learning methods for optimization and control in power grids to enable the integration of renewables and low-carbon energy.

Rankin: Why is collaboration important for tackling issues concerning AI and environmental justice?

Donti: Climate change cannot be addressed without addressing issues such as colonialism, racism, and global power structures. With this in mind, it is worth noting that the use of AI may centralize power among those select institutions and entities with the knowledge and resources to develop and deploy it, widening the digital divide. It is important that AI be leveraged in ways that address, rather than widen, societal inequities, for example through the choice of problems and stakeholders addressed. In addition, work at the intersection of AI and climate is highly multi-sectoral and interdisciplinary, requiring expertise from across academia, industry, government, and NGOs, and from both AI and climate change-related fields. It is extremely important that projects be co-developed with multi-disciplinary, multi-sectoral teams, made up of a diverse and equitable set of stakeholders who understand the problems at hand. This can help ensure that projects are well-scoped, find a clear pathway to impact, are mindful to avoid negative side effects and unintended consequences, and center and address (rather than exacerbate) equity-related issues.

Rankin: What are some ways AI can support environmental justice?

Donti: Climate action requires local solutions at a global scale. Therefore, many different applications of AI for climate action can support justice if they are developed from the ground-up alongside stakeholders from underserved contexts. This requires re-evaluation of how we fund and incentivize work in AI for climate action, noting that those stakeholders with the biggest pockets often fund AI-for-climate work and therefore disproportionately drive priorities. For instance, Climate Change AI's Innovation Grants program (which funds projects using AI for climate action) explicitly assesses the equity-related implications of proposed projects. Leveraging AI in an environmentally just manner also requires large-scale capacity-building and upskilling to democratize relevant knowledge and expertise and enable a larger set of stakeholders to develop, shape, and own AI-for-climate solutions.

One specific way AI can support environmental justice is by democratizing information availability and access across countries and contexts. For instance, by leveraging openly-available data streams such as satellite and aerial imagery to infer decision-relevant information (such as facility-level pollution and greenhouse gas emissions, crop types and crop yields, and the energy efficiency characteristics of buildings) or providing better forecasts of quantities such as weather both locally and at a global scale, which is important for both farming and disaster response.

Rankin: If you had one wish, what would you use it to fix, as it relates to AI and the environment?

Donti: I would first wish for three more wishes. :-)

The first is to equip a range of stakeholders with literacy regarding the opportunities and pitfalls associated with AI in the context of climate change, as well as the multidisciplinary skillsets (including AI expertise, climate-related domain expertise, and expertise in responsible, ethical, and participatory design) that are necessary to design, execute, and evaluate projects in a responsible, equitable, and impactful manner. Ultimately, people are the experts in their own problems and pain points; therefore, appropriate education and capacity-building has the potential to enable large-scale, ground-up action.

The second is to improve the availability of high-quality public data (especially in underserved contexts) and accessible compute. This would help remove critical practical bottlenecks to leveraging AI for climate action.

The third, and actually first in priority order, is to fully remove the negative impacts of AI on the environment. This includes both AI’s direct environmental footprint associated with its computations and hardware, and the applications of AI across society that impede climate action, for example AI for oil and gas exploration and extraction, AI for facilitating wasteful consumption, and the ways AI promotes the spread of climate misinformation.

Rankin: The AI Intersections Database can help with your first wish, so we’re off to a good start!

This post is part of a series that explores how AI impacts communities in partnership with people who appear in our AI Intersections Database (AIIDB). The AIIDB maps the spaces where social justice areas collide with AI impacts, and catalogs the people and organizations working at those intersections. Visit to learn more about the intersection of AI and Environmental Justice.