When GenAI writes public foresight reports: Effects on convincing citizens

This study investigates the impact of authorship on the convincingness and willingness to read government foresight reports. Using a preregistered online-based experimental design with 307 participants, this research examines how perceived and actual authorship (human experts vs. generative AI) influence these factors. Our findings show a significant interaction between perceived and actual authorship. Reports perceived as written by AI but actually authored by experts were less convincing, while those perceived as written by experts were more convincing. Higher perceived convincingness correlated with greater willingness to read the full text. Hence, the present research highlights biases and trust issues in AI-generated content, emphasizing the importance of aligning perceived and actual authorship to enhance government communications and public policy.