Working Papers

This paper proposes the use of synthetic training data generated by large language models
to improve machine learning SDG classifiers. It shows that supplementing existing training data with
synthetic data produced by the ChatGPT tool improves the performance of the SDGClassy classifier.
This addition of synthetic data is especially useful in building SDG classifiers given the limited availability
of properly labeled data and the complex, interconnected nature of the SDGs. Synthetic data thus enables
more effective machine-learning applications in this context.

To achieve the greatest possible human welfare, the Stockholm Environment Institute’s Climate and Regional Economics of Development (CRED) model calls for rapid reduction of greenhouse gas emissions to keep cumulative 21st century carbon dioxide emissions below 2,000 Gt. We explain why as some other models claim very slow emission reductions are best. We make three changes to the basic assumptions of the well-known DICE model to include the most recent estimates of economic damages from climate change, express greater concern about the well-being of future generations, and expect rich countries to invest in emissions and poverty reduction in poorer countries.