Brussels – Using machine learning and artificial intelligence to test how European funding for development aid affects recipient states: yesterday, CeSpi (Center for International Policy Studies) Director Daniele Frigeri and MEP Udo Bullmann presented the study carried out for the European Parliamentary Group of Socialists and Democrats (S&D) to have accurate and impartial data on the effects of European spending in this area.
The EU finances projects worth billions of euros in third countries through the Commission and the EIB (European Investment Bank) every year, aiming to help them grow or improve living conditions: poverty, health, education, gender equality, and the green transition are areas in which the EU invests. It provides funding for each project to help achieve the intended goals. However, sometimes, the money the EU disburses produces results that are not closely related to the objectives for which it was granted.
Bullman explains: “The aim was to develop a methodology enabling the European Parliament to fulfil its scrutiny role towards the European Commission and to propose improvements ahead of the NDICI (Neighbourhood, Development and International Cooperation Instrument) mid-term review”.
Marco Zupi, who coordinated the team and edited the report, with the collaboration of Christian Morabito claims that “by integrating Machine Learning (ML), Artificial Intelligence (AI), and qualitative methodologies, CeSPI aimed to dissect and comprehend EC’s actions within NDICI, their coherence with Sustainable Development Goals (SDGs), and alignment with the Socialists and Democrats (S&D) Group’s priorities”.
“This innovative approach – Zupi continues – not only modernizes scrutiny of EU development policies but also provides MEPs with a robust framework to evaluate NDICI Annual Action Plans (AAPs). CeSPI’s methodology facilitates informed and impactful policymaking, ensuring alignment with progressive priorities and values”.
The study scrutinized 176 AAPs, guiding EC’s actions across 89 countries, and conducted a detailed analysis in El Salvador, Jordan, Kenya, Nigeria, and Senegal. Despite AAPs’ professed alignment with SDGs such as human development, poverty reduction, and gender equality, the analysis reveals a discrepancy between professed goals and actual initiatives.
Recommendations include specifying targets, enhancing coordination, implementing poverty-reduction targets, collaborating for transparency, creating a High-Level Panel, and encouraging further research using TM and ML techniques for effective scrutiny.
CeSpi’s study shows that the funding area where the actual purpose differs from the planned purpose is most often in education and gender equality. The reason is that they are broad themes and involve various types of possible interventions. For example, if EU money allocated for education is to build schools, its direct impact is on construction, less on education. To test this new tool, CeSpi inserted all available funding documentation for 2021-2023 into the machine. These are exorbitant numbers of papers, thousands of documents, that a person would never be able to verify, but artificial intelligence can. Specifically, The Italian think tank studied the data by global macro areas and state-by-state.
However, a tool like this would benefit the S&D group, which could strengthen its case when negotiating for funds with the Commission, NGOs, and even the Commission, which would have greater control over its funding.
Although it is an artificial intelligence-based tool, human support remains crucial. CeSpi pointed out that to implement the operation of this tool, it would need all documents written in a unique format. Another weakness concerns language: for now, the machine can only read documents in English, but as it is still in the early experimental stages, this limitation could be overcome.
English version by the Translation Service of Withub
