Reflections on the problems and challenges experienced by countries in the implementation of SDGs as reported in the Voluntary National Reviews, Dr Petru Dumitriu
Ambassador (Ret.) Dr Petru Dumitriu
DiploFoundation
Switzerland
PURPOSE: The research aimed to accomplish a double objective: 1. To identify the main challenges and problems experienced by a sample of seven countries from four continents, in the implementation of Sustainable Development Goals (as a case study). 2. To use an AI tool for the extraction and relevant data from Voluntary National Reviews (VNRs) (as a learning experience).
DESIGN/METHODOLOGY/APPROACH: The research methodology was conceived in five steps.
1. Prompting the AI assistant: documentation and questions. 2. Overview of the first round of answers.
3. Repeating the question at a subsequent stage. 4. Overview of the second round of answers and preliminary conclusions. 5. General conclusions.
FINDINGS: The AI tool used for the purpose of this study was strictly limited to data processing and text generation. The first round of findings by the AI assistant were basically too general, vague, and almost useless. The quality of the results improved when the AI assistant was asked the same question for the second time. Even so, the findings were not convincing. However, the AI assistant helped the author to extract information from thousands of pages and may be continued to be used for that purpose.
ORIGINALITY: A concrete and transparent use of AI replaced the general and abstract talks about the future of AI, which stir up imagination but are often not based on real understanding and knowledge. The development of AI is the work of industry; researchers should not only speculate about AI’s future but try to find the best use of existing available AI tools. This research represented a good start of a learning exercise and identified a direction that may help the United Nations and its member countries in the assessment of the implementation of SDGs.
IMPLICATIONS: The results obtained so far are not satisfactory. We should not rely blindly on any AI findings, even for matters of text generation. The experiment proved that prompting AI and machine learning can improve the performance of the AI system. In both scenarios the oversight and monitoring by the human factor should remain fundamental.