



A 2026 Pew study found nearly a quarter of U.S. adults use AI chatbots daily, while another study found 48% of chatbot answers about the news contained errors. University librarians offer four steps for research: pick the right tool for the task, ask follow-up questions, trace claims back to original sources, and watch for invisible bias in what the AI leaves out.
The librarians note that tools like Claude draw from a mix described by Anthropic only in general terms, and that most AI tools favor English-language web content, potentially missing expert knowledge outside that scope. Their bottom line: the researcher, not the chatbot, remains responsible for deciding what to trust.
The full dispatch is available from the source below.