Unclear effects of AI-generated content infiltrating public and private communication

Ascertain the effects of AI-generated content infiltrating parliamentary speeches, song lyrics, movie scripts, spoken language, and interpersonal messages on public discourse, cultural production, and everyday communication practices.

Background

The authors note growing evidence that AI-generated content is already present across multiple communicative settings—from legislative speech to entertainment and everyday messaging. However, they explicitly acknowledge that the consequences of this infiltration have not been established.

This open question connects to broader concerns raised in the paper about homogenization, semantic shifts, and the potential for LLMs to subtly steer opinions and evaluative criteria.

References

While evidence has surfaced that AI-generated content is already infiltrating parliamentary speeches, song lyrics, movie scripts, spoken language, and even messages we send to our coworkers and loved ones, it is not yet clear what effects this will have.

— How LLMs Distort Our Written Language  (2603.18161 - Abdulhai et al., 18 Mar 2026) in Section 6 (Discussion)

Whether similar effects emerge in other communication contexts (e.g., peer-to-peer emails, informal messages, or non-email genres) remains an open question for future work.

— AI-Generated Email Drafts Shift Culturally Distinctive Communication Styles in Professional Email  (2609.26403 - Sakai et al., 22 Sep 2026) in Section 5, Limitations and Future Work

Whether the patterns we observe generalize to other cultural groups at intermediate points on the high/low-context spectrum remains an open question.

— AI-Generated Email Drafts Shift Culturally Distinctive Communication Styles in Professional Email  (2609.26403 - Sakai et al., 22 Sep 2026) in Section 5, Limitations and Future Work

Guo et al. (2024) demonstrate declining lexical, syntactic, and semantic diversity under recursive training on synthetic text. This provides evidence for a model-level feedback process, but its relationship to community language practices remains an empirical question.

— From Algorithmic Marginalization to AI-Mediated Re-Centering: Can Culturally Grounded AI Bring Hakka Language and Culture Back into Mainstream Society?  (2610.07672 - Chang, 6 Oct 2026) in Section 2.3, “Homogenization and recursive feedback”