A study published in PLOS One found that public participants rated AI-generated debate responses higher than actual responses from political figures on BBC1’s Question Time. Researchers used GPT-4 Turbo to generate impersonated responses based on Wikipedia biographies of 112 public figures, then had 948 UK adults evaluate both original and AI-created responses.
The findings showed statistically significant differences across all measures. “AI-generated responses were consistently rated as more authentic, coherent, and relevant than real ones,” according to the research. Notably, participants perceived AI responses as better addressing the questions posed, even when linguistic surface features appeared comparable.
Key findings
AI responses used broader vocabulary and fewer epistemic markers like “I think.” Approximately half of responses showed substantive content differences. Participants overwhelmingly wanted transparency about AI use and training methods.
The research raises concerns about misinformation potential. Prof. Herbold, of the University of Passau, stated: “This shows the enormous misinformation potential of AI that society must be aware of to critically judge any written information.”
Journal: PLOS One
DOI: 10.1371/journal.pone.0347757
Article Title: LLM-impersonated debate contributions are more authentic, relevant and coherent than their original: A representative study using BBC1’s Question Time
Publication Date: 1-Jul-2026
Institution: University of Passau
Source: EurekAlert