How AI can drive sustainability by revolutionising data sourcing

The challenge

We live in a world of ever more abundant data – and yet data limitations have major consequences for the global population and for the world. At present a lack of vital data is making it harder for policymakers to achieve the UN’s Sustainable Development Goals (SDGs) that are essential to the future health of the planet and the world. Indeed, according to the ETH university in Zurich, none of the SDGs is currently fully monitored at the global level. And, in the worst case – Goal 16: Peace, Justice, and Strong Institutions – data fulfilment from countries around the globe is at only 23.2%.

This matters a lot. During a drought how can a government plan an adequate quantity of supplies without knowing the demography in defined areas? Or how can an international organisation measure the environmental consequences of illegal timber logging without knowing the status of the forest? In these cases, reduced data availability can be detrimental to decision-makers, obstruct achievement of the SDGs, and even cost lives.

One of the main data availability challenges relates to the operation of gathering data. Indeed, in some cases, data may be too time-consuming, difficult, or even dangerous to gather, especially in remote or conflict-affected areas. Data sourcing is also a time and resource-consuming activity in which different quality and quantity issues can arise. For some topics, such as demographic counting in remote areas, data is usually accounted for manually which requires substantial human resources.

The role AI can play

To tackle this challenge various organisations are now turning to artificial intelligence (AI) to improve the quality and quantity of data sourcing. By making data sourcing tasks automatic and offering new data points, AI has the potential to help organisations focus their attention on geographical areas or population that were inaccessible, and thus to channel resource allocation to the right place at the right time. “All in on AI” provides examples of how government investments in AI for data sourcing have been helpful. For example, the National Oceanic and Atmospheric Administration in the U.S. is now using video from onshore cameras and AI to monitor fish stocks in real time. Other governments are using AI and new data points to predict natural disasters such as flooding and volcanic eruptions. AI adoption is in its infancy, but the technology is already showing the potential to accelerate data collection and help the world achieve the SDGs.

Deloitte’s Geneva workshop

To identify the role of AI in the sourcing of SDGs data Deloitte Switzerland organised a workshop in collaboration with the SDG Lab and Graduate Institute in Geneva in March 2023. Several UN agencies and non-governmental organisations active in development and the humanitarian sector participated. Two use cases of AI for data sourcing were presented by the International Committee of the Red Cross (ICRC). The first example focused on using satellite images to map populations in low-income countries. The second aimed at using text mining to monitor patterns of violence in different regions. With these new data points ICRC can better plan its humanitarian missions and thus positively impact fulfilment of Target 1 of SDG 16, which is to: “Significantly reduce all forms of violence and related death rates everywhere”.

An important challenge highlighted in the workshop was the lack of quality or validation of their data set. For climate-related information, for example, a recent global Deloitte survey found that only 5% of 130 surveyed firms think the data they use to measure, track, and achieve their net zero objectives is fully accurate or complete. Given the major implications of climate change, this is an extremely small number and could lead to un-proper climate-mitigation activities. In this specific case the use of AI can be a game changer. Firstly, automatising the process of data gathering can reduce the risks of mistakes in data entry and drive standardisation. Secondly, AI can support data validation by implementing techniques such as automatic data cleansing and formatting.

A second challenge is limited standardisation of data sources. This issue is particularly relevant, given that the SDGs have extensive synergies and trade-offs associated with their development. The UN mentions the possible trade-off between the improvement of Goal 2: Zero Hunger, and the related, possibly detrimental impacts on Goal 15: Life of Land. There is, on the other hand, a positive synergy with Goal 10: Reduction in Inequalities.

As synergies and trade-offs are so common, they are critical in creating standards for comparisons. AI can support this standardisation and provide a new range of data sources for international organisations, such as satellite imaging, social media insights, GPS data, video, and images, and many more. For example, policymakers tackling food and nutrition questions could start using climate-related data to better plan for drought and hunger.

A third challenge discussed was the lack of quantity of valuable data to drive meaningful decision-making. As mentioned above, most international organisations lack access to the necessary data for their operations. New data sources can help mitigate this issue. For example, the use of satellite images can offer a plethora of data for organisations monitoring deforestation or poaching: manual image processing is too time consuming; AI has the potential to streamline the task.

Good data practices to get started

As this shows, AI can help significantly to solve data-sourcing issues for the SDGs. But establishing AI for data sourcing is not straightforward and requires time and investment. There are, however, good practices that can help organisations get started. We recommend the following:

• Incentives to process data – The data provider needs to see value in the data they are sharing. Organisations need to find ways to motivate citizens and communities to share their data: for example, by convincing communities to provide data that can improve the biodiversity and quality of their forest and thereby gain access to carbon credit markets.
• Do not harm - Organisations should also consider the purpose of their data gathering, to ensure that it’s done for proper and transparent reasons. AI ethics is important, specifically when considering the social areas of the SDGs such as diversity and health. Being transparent about how and why the data will be used and stored should instil the necessary confidence in stakeholders along the value chain to provide quality data.
• Foster ecosystem creation – Sustainability reaches across countries and industries. Many actors are generating data through their activities, that could prove valuable to other organisations, but data sourcing and processing is often not shared. Transparency and collaboration can generate value for all the actors in the ecosystem. For example, organisations working in healthcare can collaborate to establish health data standards that can easily be shared by governments to help create the data to support them in their mission.

Lack of data to report progress towards the SDGs and broader sustainability goals is a significant concern that AI can help to mitigate by providing new data points and higher quality information. The challenges and solutions identified during the workshop and highlighted above are not exhaustive or conclusive, but they provided a starting point for continued collaboration and innovation. Further events about the potential of AI will be held with Graduate Institute and the SDG Lab in the coming year to establish a community and drive collaboration. Reach out if you are interested to learn more.

Thank you to Deborah de Wolff for her valuable contribution to this article.