3 strategies to leverage AI in the development sector

More than $200 billion per year is currently being spent in the development sector to support around 300,000 development projects in 160 countries. Development projects fulfil a variety of purposes. Examples include: building schools to give children access to education, providing medication to improve healthcare, or installing solar panels to increase access to electricity.

The common agenda for all development projects is to contribute to the United Nations’ Sustainable Development Goals (SDGs) and thus promote a better future for the people and the planet.

A salient challenge for decision-makers in the development sector, however, is to effectively align the myriad of global development projects to the needs of countries and regions. To achieve this, decision-makers in the development sector need evidence-based decision support. So far, evidence is mostly generated manually without tools, for example, via surveys, which are expensive and further limited in coverage and granularity.

Only recently, the use of artificial intelligence (AI) has gained momentum in informing policy decisions in the development sector. This opens up entirely new possibilities: AI solutions can locate households with low-income levels, with no access to electricity, or at risk of food shortages, with a surprisingly high degree of accuracy and at a low cost, thereby enabling better decision-making.

Over the last few years, several proof-of-concept projects have used AI to support decision-makers in the development sector. For example, AI has been used to map poverty via mobile phone metadata. The AI was later used during the COVID-19 pandemic to identify people in need and transfer cash directly to their phones, thus avoiding the usual bureaucratic bottlenecks. AI has also shown great potential to measure progress towards the SDGs by analyzing large-scale satellite imagery. One spin-off, Atlas AI, uses this technology to incentivize capital investment in emerging markets by identifying underserved markets and regions.

Other projects with AI technologies focus on the supply side of development finance. For example, AI technologies using natural language processing are able to analyze millions of written reports from development projects, which, in turn, helps to better coordinate development projects across countries and SDGs and also enables independent verification of international climate finance flows.

Besides a few proof-of-concept projects, actual AI implementations in the development sector are still rare. Based on our own experience from collaborating with development organizations, we see three prime reasons: First, internal resources and competencies around AI must still be developed, thereby hampering innovation processes. Second, the development sector is – in principle – highly attractive for AI talents who aim to make a meaningful impact. However, few AI talents have so far found their way into the development sector and often move towards roles in industry. Third, AI is also changing how decisions are made, requiring new processes and organizational structures.

However, revamping existing decision-making structures in organizations is challenging, especially when integrating new technologies like AI.

By bringing AI into the development sector, new evidence can be generated for better decision-making. To achieve this, three strategies should be key priorities:

1. Promoting external development: Since many development organizations lack internal resources and competencies to develop new AI innovations, it will be crucial for them to leverage external competencies through collaboration. Here, the paradigm of “open innovation” can be a cost-effective way to develop new solutions and bring them into established development organizations. For this, development organizations should connect to various stakeholders, including startups, non-governmental organizations, and academic institutions.

While open innovation is nowadays common practice in companies, similar efforts in development organizations are still rare. One promising exception is the International Committee of the Red Cross (ICRC), which partnered with universities to translate new AI technologies into humanitarian action through close collaborations. Such collaborations can harness the respective skillsets of stakeholders: development organizations can focus on their unique domain knowledge while external collaborators bring expertise for developing and implementing AI solutions.

2. Building up AI competencies: To implement and maintain AI solutions in the long run, development organizations need to build up internal AI competencies. For this, organizations need to find a balance between upskilling employees and extending their recruitment to attract top AI talents. This will require decision-makers to define new organizational roles (e.g., AI engineers) equipped with long-term, expert-focused career paths. Notwithstanding, the current competition around AI talents is hard. However, development organizations are in a unique position to offer meaningful work tasks that can generate substantial impact around the world.

One powerful means to support this process is through hackathons where interdisciplinary teams work on real-world AI challenges from the development sector. This exposes existing employees to AI-centered problem-solving, while AI talents become aware of the unique challenges from the development sector and how they can contribute with their skills to develop impactful solutions. For example, during “Hack4Good” hackathons, interdisciplinary teams spend two months working on data science challenges provided by organizations such as the World Wildlife Fund (WWF) and the German Agency for International Cooperation (GIZ). This way, internal AI competencies can be developed while also attracting new AI talents from outside.

3. Creating an entrepreneurial culture: Organizational resistance to change is one reason why organizations often fail to embrace new technologies such as AI. This holds especially true for “bureaucratic” organizations in the development sector with large interdependencies and long-ranging decision-making processes that may “kill” innovative ideas early on. Many companies have faced similar problems and, therefore, create organizational sub-structures that allow for more dynamic processes as well as fast piloting and scaling of ideas.

Decision-makers should adopt similar practices. Depending on the organization, different ways can be of use. One way is to create separate, startup-like entities within an organization. For example, the Organization of Economic Cooperation and Development (OECD) launched the SDG Financing Lab as a new vehicle to bundle first-mover activities around AI. This eventually resulted in the first-ever AI solution at the OECD: a dashboard for monitoring contributions to the SDGs through global development aid.

Another vehicle is accelerator programmes that directly connect startups with development organizations. Such accelerator programmes allow startups to keep the benefits from their small size and organizational independence while providing them with the benefits of a large organization through mentoring and, more importantly, resources (e.g., data, personnel, finances, access to operations). For example, the World Food Programme has launched the WFP Innovation Accelerator, which supports young startups in combating world hunger by giving them access to its global operations. As a result, several new AI-based technologies have been piloted, such as the HungerMap Live to monitor global food security in real-time and Optimus, which guides the World Food Programme in a cost-effective design of food baskets and their subsequent delivery.

The above strategies should be key priorities for decision-makers to “act now” in order to leverage the potential of AI. Besides them, it is important that decision-makers in the development sector set a long-term vision with a coherent set of strategies for how integrating AI into their organizations. These strategies should be regularly updated and complemented over time. Here, early success stories can further help in overcoming organizational resistance by convincing employees and other stakeholders of the merits of AI, and eventually, help in scaling AI technologies solutions across entire organizations.

In recent years, AI has triggered a fundamental transformation of decision-making in industry, and AI is now about to unfold its impact in the development sector and aid public leaders through evidence-based decision support. In the near future, we will see that AI will become an integral part in the decision-making of development organizations. As a result, new AI technologies will make a substantial contribution towards reaching the SDGs by 2030.