AI-Generated Email Drafts Shift Culturally Distinctive Communication Styles in Professional Email

This study investigates whether AI writing assistants, which often default to Western-centric communication norms, influence the culturally distinct ways professionals write emails. The researchers conducted a preregistered within-subject experiment with 175 Japanese and American participants. Each participant wrote three workplace emails (a favor-request, a disagreement, and a task-refusal) in three conditions: without AI, with a low-context AI draft, and with a high-context AI draft. The study measured the use of linguistic markers associated with indirectness and imposition mitigation to quantify the cultural communication style of the resulting emails.

The researchers found that Japanese participants naturally used significantly more high-context markers (such as politeness and apologies) than American participants in the no-AI baseline. However, when using AI, participants' writing shifted toward the style of the AI draft. Crucially, this shift was not uniform: it was significantly larger when the AI draft was culturally misaligned with the user's background. Japanese participants drifted more under low-context drafts, while American participants drifted more under high-context drafts. Despite these shifts, participants showed high overall reliance on AI drafts, retaining the majority of the AI-generated text regardless of its cultural orientation.

These findings suggest that AI writing tools act as a force for cultural homogenization, potentially overwriting distinct communication norms that are essential for maintaining professional relationships and social harmony. Because users frequently accept AI drafts with minimal editing, these biases can propagate quietly. The results highlight a critical need for AI assistants that are not designed as culturally neutral, but rather as systems capable of recognizing and adapting to the diverse communication styles of their users.