LLM-generated messages can persuade humans on policy issues.

Nat Commun · 2025 · PMC12215518 · PMID 40593786

Europe PMC DOI

hedged sentence

0.0840

closest p · 1.7× alpha

0.0840

boldest claim

The sentences

marginally significantp = 0.084so close (0.05 < p ≤ 0.1)

Accordingly, all results are robust, except that different perceptions of how “smart” the authors of the AI-generated (versus human-generated) messages were seen to be became non-significant ( p = 0.504) and “unique” became marginally significant ( p = 0.084) and therefore these results are not visualized in Fig. 2 .

also in 26,082 other papers

Quoted from the open-access full text in Europe PMC under the licence the publisher applied. The sentence is reproduced exactly as published; the emphasis is ours.