The COVID-19 pandemic created an incredibly challenging situation in the international development space. On the one hand, developing countries were going through an extraordinary crisis, which increased the global demand for international development assistance. On the other hand, COVID-19 slowed down the economies of the donor countries, which faced new socio-economic challenges at home. This article is a summary of statistical research that looks at how the Gross National Income of the Organization for Economic Co-operation and Development (OECD) member states changed between 2018 and 2021 and correlates it with the amount of money they invested in Official Development Assistance (ODA) during this period. The statistical analysis shows that despite economic challenges and increased demand for social assistance at home, OECD countries actually increased their international development assistance spending.

Artificial intelligence chatbot, ChatGPT, was first released on November 30, 2022, and that was right around the time I was working on this research. So, I registered on the ChatGPT website and asked it to write a paragraph about the impact of Covid-19 on official development assistance. Within 5 seconds, ChatGPT wrote a persuasive paragraph arguing that “many donor countries have faced economic challenges and have had to redirect funds towards domestic priorities, such as healthcare and support for businesses and individuals affected by the pandemic.” It concluded that “ODA funding has been reduced, which has had a detrimental effect on the ability of recipient countries to address development challenges and achieve their development goals.” The text was fluid, logically coherent, and not plagiarized. However, the analysis of actual data showed that despite its sound reasoning, ChatGPT was wrong.  

I collected all the data from the website of the Organization for Economic Co-operation and Development (OECD), which is an international forum that brings together 38 of the world’s most economically advanced countries to exchange best practices, tackle common problems and contribute to global peace and development. OECD countries account for 18% of the world population but 63% of the global GDP and about 95% of the Official Development Assistance. I gathered data for a four-year period from 2018 to 2021 and looked at mainly two indicators: 1. data on the Gross National Income, as it is one of the most important indicators of a country’s economic performance, and 2. ODA represents the amount of money a donor country spends on international development assistance. 

Between 2020 and 2019, the GNI of OECD countries shrank by $1,3 trillion, which means an average of $34 billion reduction per country. However, contrary to GNI, ODA spending of the OECD countries increased by $6,6 billion, which breaks down to a $193 million average per country. Consistent with these findings, the average ODA, as a percentage of GNI, increased from 0,373% in 2019 to 0,399% in 2020.      

Conclusion 1 

So, the statistical analysis showed that despite the negative impact of COVID-19 on their domestic economies, the OECD countries increased the amount of money spent on Official Development Assistance. This analysis leaves us with a positive and encouraging message that in times of need, the international community is able to come together and take action for the common good beyond national borders. 

Conclusion 2 

It is also another reminder about the limitations of large language models, such as Chat GPT. There are several generative AI models which can produce somewhat original text, images, and other data, based on a statistical analysis of information on the internet. For example, Chat GPT, the most successful of these models, was trained on 570GB of information (approximately 385 million pages on Microsoft Word), allowing it to generate human-like seamless outputs within seconds. However, due to associated costs, the model is pre-trained, which means there is a cut-off date for source information. In addition, all the misinformation, fake news, and biased texts found online are also fed into training the model. The engineers at Open AI are taking measures to tackle the issue of misinformation, but that is a tall order, even if you have the best intentions. So, while humans can investigate and find the truth, it is a much more elusive task for pre-trained AI models. 

P.S. Most of the OECD countries still need to catch up to their commitment to spend 0.7% of their GNI on development assistance.