Exploring the Impact of Artificial Intelligence-Mediated Communication on Bias and Information Loss in Non-academic and Academic Writing Contexts

AI-MC’s impact on bias and information loss is underexplored. We conducted two studies (non-academic: N=572; academic: N=420), comparing original Chinese texts with those refined by ChatGPT 4.0, Claude 3 Opus, or Gemini Advanced (non-academic only). Bias was measured via a 5-point Likert scale, and information loss via comprehension questions. Mann-Whitney U tests showed ChatGPT 4.0 reduced non-academic emotional bias with a small effect (r = .18, p < .01), while Gemini Advanced increased bias in specific cases (p < .05). No models affected information loss (p > .05) or academic bias (p > .05). This supports cautious AI use in academic editing and further cross-linguistic research.