Benchmarking Debiasing Methods for LLM-based Parameter Estimates

Nicolas Audinet de Pieuchon, Adel Daoud, Connor T. Jerzak +2

Large language models (LLMs) offer an inexpensive yet powerful way to annotate text, but are often inconsistent when compared with experts. These errors can bias downstream estimat…