Interview with Dona Mathew
"Data centres and data transmission networks contribute to 1-2 per cent of greenhouse gas emissions."

How can artificial intelligence help address the pressing challenges of climate change? In this interview with Dona Mathew, an expert at the Digital Futures Lab, we explore the various ways AI can be utilized in areas such as weather forecasting, climate-resilient agriculture, and energy efficiency. Dona shares her insights on the potential and challenges of AI, as well as the ethical considerations that should be taken into account when implementing AI in climate-related initiatives.

In what ways can AI be harnessed to effectively combat climate change and promote sustainability?

There is a growing wave of experimentation and emerging use cases of AI in areas like weather forecasting, climate-resilient agriculture, energy efficiency, and disaster response. Machine learning can help draw meaningful insights by analysing vast datasets, such as historical weather records and satellite images of Earth's topography. However, as AI in this space is still developing, its long-term effectiveness in addressing climate change remains uncertain.

Much of AI's potential seems to lie in its ability to monitor and track environmental changes. It can be used to monitor deforestation, assess flood risks, track pollution levels, and observe biodiversity loss. By processing large volumes of data, AI could potentially identify patterns in environmental degradation or predict weather events, aiding real-time decision-making.

However, these applications come with challenges. Decisions based on AI outputs must carefully consider potential negative impacts on local communities. For instance, while AI can identify optimal locations for solar or wind energy projects, decisions about land use must also account for the reliance of forest communities on those lands for food and livelihoods.

Considering AI's impact, do you believe its overall influence on global issues like climate change is more beneficial or harmful? What AI-related challenges do you see here?

It’s difficult to categorize AI as purely beneficial or harmful, especially in light of recent discussions on its environmental impact. When Digital Futures Lab conducted its study on the intersection of AI and climate action in Asia, the literature on AI’s environmental costs was only beginning to emerge. This year, however, we’ve seen significant developments with more concrete data. The findings from various measurement initiatives are concerning. For instance, generating 5-50 prompts on ChatGPT uses around 500mL of water, and generating images with advanced AI models consumes as much energy as fully charging a smartphone. On a global scale, data centres and data transmission networks contribute to 1-2% of greenhouse gas emissions.

Despite this, in early October 2024, former Google CEO Eric Schmidt suggested building more AI data centres, arguing that since humanity is already off-track on climate goals, the environmental impact of AI should be less of a concern. This downplaying of AI’s climate effects and the focus on its ‘potential’ benefits should not be allowed to distract from meaningful climate solutions that tackle the core issue: the anthropocentric nature of economic growth and development.

Of course, there are additional challenges to adopting AI, including biases and inaccuracies rooted in existing datasets, the marginalization of vulnerable groups due to the digital divide, and the profit-driven nature of AI interventions that often exclude those most at risk.

What ethical considerations should guide the deployment of AI in climate-related initiatives?

In contexts where social inequalities are pronounced, and rights enforcement is often limited, the implementation of AI for climate action must be approached with responsibility. This involves using participatory methods to identify problems and co-design technology solutions. From the outset, data collection should adhere to FAIR (Findable, Accessible, Interoperable and Reusable) and CARE (Collective Benefit, Authority to Control, Responsibility and Ethics) principles to ensure responsible climate data governance. Additionally, more equitable data-sharing mechanisms and opportunities for interdisciplinary collaboration are essential. Importantly, AI interventions should avoid creating technological dependencies or product lock-ins, ensuring sustainable solutions for the future.

The path to realising these principles is complex and demands collective action. A key starting point is ensuring that diverse stakeholder voices are heard, and that there are platforms for engagement between civil society, technologists, businesses, and communities, all working toward equitable climate futures. To promote interdisciplinary knowledge exchange and raise awareness about the intersection of AI and climate, DFL recently launched Code Green, a new media series aimed at breaking down critical questions related to this intersection for a broader audience.