Can Large Language Models Reliably Code Qualitative Humanitarian Data? A Benchmark Study Against Human Expert Adjudication

Jerome Marston, Tino Kreutzer, Salomé Garnier +3

Data from affected populations are crucial for informing humanitarian response, but their value depends on timely and consistent interpretation of nuanced accounts of need. Humanit…