Digital Futures Lab recently developed the AI and Climate Futures project to look at how AI can be leveraged to manage the climate crisis in Asia. Based on your findings, how could AI help to enhance global climate governance?
When we created the “AI + Climate Futures in Asia” initiative in 2023, little was known about how these two critical forces that shape the region – digital transformation and climate change – intersected. We set out to map the AI climate landscape across nine countries and identified applications of AI in climate-relevant sectors (“use cases”) while logging enablers for the adoption of AI and documenting challenges in the process. To overcome this clear knowledge gap, we built an expert network, foregrounding diverse backgrounds and experiences, and focused our academic exploration on three key domains: agriculture, energy, and disaster risk reduction.
AI’s ability to both process vast datasets from multiple sources and make this information actionable via machine learning makes it a useful tool for policy making. For instance, the Data in Climate Resilient Agriculture (DICRA) platform) leverages agricultural data to guide policymakers in developing climate-resilient farming practices. Another application of AI is demonstrated by the GeoAI platform, a collaboration between the United Nations Development Programme (UNDP)and the University of Nottingham. Using labelled satellite images and a computer vision algorithm, this platform monitors brick kilns in India – major hot spots for pollution – and identifies non-compliance with environmental policies. This tool not only tracks kilns but also supports coordinated action by diverse stakeholders, from civil society to regulators. It provides information on pollution and integrates context-sensitivity so as not to disrupt livelihoods in transitions to greener manufacturing processes.
In the energy sector, AI’s role in renewable energy (RE) integration stands out. Traditional power systems have predictable demand and supply patterns – variable renewable energy (VRE) like solar and wind less so. AI can forecast demand and supply, which in turn enables smoother transitions to RE and reduces emissions over time.
Considering the significant energy consumption associated with AI technologies, how can policymakers balance the benefits of AI in climate mitigation with the imperative to reduce carbon footprints?
This is indeed a complex challenge. And it is further complicated by a lack of transparency when it comes to reporting key data around electricity, water consumption, and greenhouse gas (GHG) emissions. On the upside, it is a topic that has gained significant traction since 2023. When we started our research, we quickly identified the discourse on the environmental impact of AI by using foresight methodologies (like horizon scanning), but also noticed that the conversation on AI and climate was largely confined to policy circles and academia. Since then, however, this concern has become more mainstream. Workshops and public events on AI and climate consistently highlight environmental impact as a key issue for stakeholders and curious participants alike.
Through our discussions with experts, we have come to several key points policymakers should consider when balancing AI’s potential benefits with its known negative impacts. The first issue is the lack of transparency when it comes to how much energy and water AI actually consumes. While we have general data on the resources data centres and transmission networks use, we don’t know how this corresponds to AI-specific consumption. We can deduce that generative AI is indeed demanding increasingly more energy and water by looking at big tech’s rising resource consumption – but these are merely estimates. So, one piece of advice we have for policymakers would be to approach consumption numbers with caution and support the creation of evaluation metrics for accurate readings and inferences.
To ensure that AI’s growth does not come at the cost of environmental and social equity, policymakers and AI developers alike will need to invest in greater transparency, equitable value distribution, and sustainable practices.
A second recommendation would be to differentiate between AI types and their specific resource needs. Crucially, AI systems do not all have the same environmental impact. Generative AI, for example, is far more resource-intensive than other models. Policymakers should be supported in understanding these nuances, so they can assess the suitability of diverse AI applications and make decisions around appropriate AI technologies for climate action.
Another way to reduce the environmental impact: develop AI models built for specific needs, rather than a ‘one-size-fits-all’ model. Not only would this help combat redundant resource use: a fit-for-purpose AI model is much more effective when it comes to climate science. Within agriculture, for example, variations in agro-climatic zones as well as diversity in plant and insect species require a more tailored AI intervention that uses specific datasets. An off-the-shelf model is not going to cut it.
A final element policymakers should examine is AI’s broader footprint – artificial intelligence not only claims a large quantity of energy and water, it has far-reaching socio-economic consequences, ranging from issues surrounding land use to the ethics of data work. What is more, the benefits and risks of AI are not distributed equally. AI generates significant monetary value for big tech companies in the Global North, while it requires a lot of groundwork labour (like data labelling) that is performed in the Global South – under much less favourable working conditions and oftentimes with very low wages for the local labour force.
In order to address these challenges and ensure that AI’s growth does not come at the cost of environmental and social equity, policymakers and AI developers alike will need to invest in greater transparency, equitable value distribution, and sustainable practices.
What role do international organisations (IOs) play in supporting the responsible use of AI for climate action?
There are a couple different ways IOs could approach this issue. At Digital Futures Lab, we have tried to foster dialogue towards sustainable AI use for climate action by launching Code Green, a media series featuring a newsletter and podcast that functions as a platform for multidisciplinary collaboration. Knowing that diversity is of great importance in AI climate research, the podcast aims to bring together experts from various disciplines and geographies. IOs could think about supporting similar platforms for collaboration.
One take-away from conversations on Code Green: while AI can enhance climate-related sectors, foundational scientific knowledge is indispensable. Decisions cannot rely on data alone: foundational science is critical for informed, effective action. In order to ensure responsible AI, IOs should therefore support foundational climate science research.
Only a broad, ecosystem approach to AI for climate can ensure sustainable and equitable digital futures for developing economies.
Another approach could be to incentivise “climate-first” investments rather than innovative AI just for the sake of technological novelty. While AI has much to offer when it comes to climate action, it is essential to avoid “techno-solutionism” and distract resources away from more effective climate strategies – the climate crisis must remain central to funding decisions.
Finally, IOs can promote community-driven, localised approaches to data collection as a way to circumvent the current, top-down, and extractive status quo. Vast quantities of unused data are stored in “data graveyards,” consuming energy and increasing emissions, while undermining community access and control over data. Initiatives like CoreStack are trying to change that. They collaborate with communities to manage natural resources and curate datasets on issues like groundwater management in order to empower local stakeholders to make informed decisions. These bottom-up approaches ensure that data collection directly benefits the communities involved. Only a broad, ecosystem approach to AI for climate can ensure sustainable and equitable digital futures for developing economies.
Urvashi Aneja is the Founder and Executive Director of Digital Futures Lab, a multi-disciplinary research collective studying the societal impacts of technology transitions in India and the Majority Word.
You can also find content from ENSURED's Expert Blog on the Global Policy website.
Photo: Nidia Dias / Pexel (via CC BY-NC-ND 2.0)
.png)
