A study published on Wednesday in PLOS One found that AI chatbots prompted to impersonate 112 public figures produced responses perceived as more authentic, coherent, and relevant than the real thing, underscoring “a dire need to inform the general public of the potential harm this can have on society.” Researchers asked GPT-4 Turbo to impersonate the public figures during the lead-up to the 2024 election in the United Kingdom, training it on Question Time, a BBC One show. The dataset included 112 speakers: politicians, business people, journalists, medical experts, writers, and “other well-known members of UK society.” After additional prompting with Wikipedia biographies, the AI generated responses to audience questions. A representative sample of 948 participants in the UK rated the AI-generated content as more authentic, coherent, and relevant than actual debate responses, showing LLMs “can be made to deceive the public regarding the nature of statements in the political domain.” Steffen Herbold, a professor at the University of Passau who led the study, said the high authenticity ratings were “really surprising because that's supposedly hard to fake,” noting the politicians were from “one of the biggest shows in the UK.” Herbold added that the real politicians spoke off the cuff, whereas the LLM drew from pre-existing text.