Bayesian Experimental Design with Large Language Models (BED-LLM) was jointly proposed by Apple, Oxford University, and City University of Hong Kong in August 2025. The relevant research results were published in the paper "BED-LLM: Intelligent Information Gathering with LLMs and Bayesian Experimental Design".

BED-LLM iteratively selects questions to maximize the expected information gain (EIG) on the task of interest. Compared to direct prompt LLM and other adaptive design strategies, BED-LLM achieves significant performance improvements in multiple tests based on a 20-question game by actively inferring user preferences using LLM.