Popular large language models (LLMs) are unable to provide reliable information about key public services such as health, taxes and benefits, the Open Data Institute (ODI) has found. Drawing on more ...