How AI is transforming climate risk management in developing countries

Typhoon Yagi and Storm Boris in September 2024, and Hurricane Milton in October 2024 illustrate the mounting climate challenges we are faced with these days. Developing countries stand on the front lines of these extreme weather events as climate change is expected to affect them more than the developed world. Only in the past few months we have witnessed numerous catastrophic flooding events across Africa and Asia. 

During Risk Awareness Week 2024, I had the opportunity to talk about how artificial intelligence (AI), with its ability to process vast amounts of data and make complex predictions, offers a transformative toolset to address climate risks. In this article, I'd like to build on what Alex Hong, my fellow Illuminem Thought Leader, wrote in a 2023 article about the benefits of integrating AI and machine learning (ML) solutions for sustainability and climate change risk management. Hong focused on the use of these technologies for businesses in ASEAN. Here, I'd like to highlight several potential advantages of these technologies for a wider array of entities as well, especially in developing countries, in the context of climate change risk assessments. 

In recent years, the introduction of AI and ML has opened new opportunities for resilience-building, allowing risk managers to use more accurate climate models that utilize early warning systems, sensor data and satellite imagery. In fact, from a risk manager's perspective, there is no doubt AI and automation will replace, in a few years' time, many of the daily risk management tasks. In this context, this article explores how AI can help alleviate climate risks especially in developing countries, addressing existing bottlenecks and opening new pathways toward climate resilience.

AI offers a transformative toolset to address climate risks, especially in developing countries, where traditional infrastructure is lacking

AI: a game changer for climate risk management

Observant or knowledgeable readers would agree that AI is already being deployed across the world for tackling climate change, in general, and for conducting climate risk assessments, in particular. There are many data and service providers, especially in OECD countries which started to use AI to manage climate risks. First, there are many available open-sourced data suppliers, such as CLIMADA, EarthScan, OS-Climate, Copernicus Climate Change Service (C3S), RiskScape, etc. which allow people around the world to access climate-related data. Alongside these, there are also many commercial platforms like Jupiter Intelligence, Tomorrow.io, Riskthinking.ai, Munich RE's NatCatSERVICE and LRI, which are all important sources of information and software solutions for different climate analyses. 

AI is already being utilized across the Global South to tackle a range of climate-related challenges. For example, Google’s AI-powered Flood Hub predicts floods and provides early warnings in 80 countries, 23 of which are in Africa. Another initiative from Google is the Open Buildings platform, an open-access dataset that maps building locations, sizes, and densities using machine learning and satellite imagery. This platform supports urban planning, emergency response, infrastructure development, and environmental monitoring. Likewise, Microsoft’s AI for Good Lab uses geospatial models to improve crop yields and promote sustainable water use in Northern Kenya. Additionally, IBM, in collaboration with the African Union and local governments, employs AI and hybrid cloud models to strengthen climate risk modeling across Africa. In areas with limited data and vulnerable infrastructure, these AI tools offer valuable predictive insights to help mitigate the effects of floods, wildfires, and other natural disasters.

Bottlenecks in climate risk assessments in developing countries

Despite the promises of these innovations, developing countries encounter significant challenges that obstruct effective climate risk assessment in general, let alone, conduct these assessments with the use of AI and ML. Key obstacles include the scarcity of reliable data, limited access to advanced technology, and political barriers that slow the progress of climate adaptation strategies.

Data scarcity and quality: In many developing countries, the lack of sufficient, reliable, and high-quality environmental data poses a major challenge. Often, these nations depend on global datasets, which frequently lack the granularity needed for accurate climate modeling at the local level. This issue is particularly critical when assessing the elements at risk (exposure), damage functions, and vulnerability. The situation is further worsened by the limited number of weather stations and data-collection infrastructure in many areas.

Technological and financial constraints: Organizations in developing countries are frequently unable to adopt advanced technologies due to a lack of funding and technical know-how. Even though data from commercial platforms can be incredibly valuable for practitioners who have less education or experience in risk management, many regions lack the resources and skills needed to fully leverage these tools. Addressing the funding and capacity gaps is crucial to enabling broader access. Additionally, although AI has the potential to help tackle climate risks, many developing countries don’t have the necessary infrastructure - such as cloud computing, reliable internet, or even consistent electricity - to effectively implement these solutions, making it harder to tap into AI-driven climate adaptation efforts.

Political and governance barriers: Political instability and competing interests can significantly hinder the implementation of effective climate risk management strategies. In many instances, resources are misallocated due to political priorities that do not align with the most vulnerable communities’ needs. Additionally, governance structures may lack either the capacity or the willingness to incorporate AI-driven climate risk tools into their decision-making process. I personally witnessed a situation where authorities were unwilling to make use of accessible and cheap technologies to monitor environmental pollutants for the well-being and health of their communities. We can only speculate as to why.

How AI helps overcome these bottlenecks

AI's potential to address these challenges is remarkable. By providing better data, improving access to climate information, and offering advanced risk assessment tools, AI can help governments, communities, and even individuals in developing countries overcome some of the key bottlenecks to effective climate risk management and can help bypass those who are not willing to make the data transparent and available.

By relying on data-driven insights, AI reduces human biases and fosters more objective decision-making in climate risk management.

1. Enhanced data collection, predictive analytics and reduced human biases

AI excels in managing large datasets and in identifying patterns that are difficult to discern manually. In regions with limited ground-based data collection infrastructure, AI tools can utilize satellite imagery, drone data, and sensor networks to fill in critical gaps. For example, IBM’s flood detection system (mentioned above) leverages satellite data to generate real-time flood extent maps, helping African governments anticipate and mitigate flood damage.

Moreover, AI models can be used to predict climate events such as storms, floods and droughts with increasing accuracy. These predictive capabilities along with incorporating solutions such as early warning systems, allow communities to prepare in advance and reduce the loss of life and property. In the context of climate change, where historical patterns are often disrupted, AI-driven models offer more dynamic and responsive solutions to adapt to emerging climate risks.

This point leads me to the issue of human biases. Like in any other field, AI plays a crucial role in minimizing cognitive biases that can distort risk management decisions. In climate risk identification, AI eliminates confirmation and availability biases, where managers might focus only on familiar or recent risks, by analyzing a broader range of data without prejudice. During risk assessment, AI allows us to avoid anchoring on initial data or overemphasizing negative outcomes. This ensures a balanced evaluation of risks. AI would allow a more scientific approach to the weight selection of criteria. As for risk mitigation, AI can prevent a common problem especially in developing countries - conservatism bias - which favors outdated methods. By relying on data-driven insights, AI fosters more objective and effective decision-making throughout the climate risk management process.

AI can break down political barriers, ensuring that climate risk assessments prioritize community needs over political agendas.

2. Democratization of climate risk management

The potential of AI to democratize access to climate risk management tools is, in my opinion, one of its most significant effects. These days, crucial insights on infrastructure risks are offered by open-source AI technologies. This could help developing nations' most isolated or underdeveloped areas make well-informed adaptation decisions. AI provides objective, data-driven insights that are less vulnerable to political manipulation, which helps to manage conflicts where there are severe tensions between community demands and government interests. Furthermore, AI models can prioritize action based on the real needs of communities rather than political objectives or the private interests of a select few by lowering reliance on subjective risk assessments or on assessments that lack reliable climate data.

A 2023 policy report by SIPRI highlights how AI can support climate security by helping policymakers and local communities make better decisions using new data sources. AI can also overcome traditional infrastructure challenges. In cases where political leaders focus on large projects that only benefit a few, or when governments don’t prioritize climate adaptation, local organizations can use AI to assess risks and implement strategies that help vulnerable communities. This approach shifts, or decentralizes, climate risk management into the hands of those directly affected, giving them more control over their future.

3. Holistic solutions for complex risks

AI provides a more integrated approach to climate risk management, addressing the silos often found in traditional risk assessments. Most conventional models look at one hazard at a time, like floods or droughts, and don’t account for how different events are linked. This is because climate risk assessments involve complex data, high uncertainty, and large amounts of information. AI, however, can combine different types of data - such as environmental, socio-economic, and infrastructure factors - to create more comprehensive and holistic risk assessments that consider multiple, interconnected risks and that consider their cascading effects. Nevertheless, these assessments must meet certain performance standards, like accuracy and validation tests, before they can be applied in real-world situations.