The Global School Forum presents: Can Machine Learning (NLP, LLM) Make Policy Research more “Reflexive”?

This talk will discuss the potential of making policy research more “reflexive” by combining qualitative and quantitative methods and employing Natural Language Processing (NLP) methods and LLMs to reduce the distance between researchers and the people they study.  This can substantially increase the relevance of research by bringing it closer to lived realities and thus improve policy effectiveness. The talk will provide a general overview of this idea and illustrate it with two examples: (1) On deliberative democracy, and (2) On understanding human aspirations.  It will also show that there are pitfalls from over-relying on NLP and LLMs without critical human input.

Vijayendra Rao is a Lead Economist in the Development Research Group of the World Bank who works at the intersection of scholarship and practice. He integrates his training in economics with theories and methods from anthropology, sociology and political science to study the social, cultural, and political context of extreme poverty in developing countries.  His books include Oral Democracy, Localizing Development: Does Participation Work?, and Culture and Public Action (edited).   His articles have been published in the leading journals in Economics, Political Science, Sociology and Development Studies. His policy work has focused on issues of bottom-up participation, local democracy and gender. He is Chair of the Advisory Committee of the program on Boundaries, Membership and Belonging at the Canadian Institute for Advanced Research.