WildChat / research hub

Uses WildChat

Perceived Trust and Transparency in AI-Altered vs. Human-Written LinkedIn Posts

Thomas Christian Joos, Azza Bouleimen, Anikó Hannák, Nicolò Pagan

arXiv preprint · First arXiv submission

Uses real AI-assisted writing requests from WildChat to design alteration conditions for LinkedIn-style posts. Compares readers’ judgments of human-written posts and versions altered by AI.

Data used

WildChat conversations are analyzed to derive realistic social-media rewriting prompts; a separate preregistered experiment involves 214 participants.

Findings

The experiment detects no overall effect of AI alteration on perceived trustworthiness or transparency. Participants identify AI alteration with 49% accuracy; ratings vary across prompts, models, and posts.

Section 3.1: WildChat formative analysis. Subsequent methods and results: preregistered within-subject experiment. Cites the original WildChat paper. Read the source.

Dates and versions

First arXiv version. ArXiv version history.

arXiv preprint.

Cite this paper

Thomas Christian Joos, Azza Bouleimen, Anikó Hannák, Nicolò Pagan. 2026. Perceived Trust and Transparency in AI-Altered vs. Human-Written LinkedIn Posts. arXiv preprint.

Show BibTeX
@misc{aialteredlinkedin2026,
  title = {{Perceived Trust and Transparency in AI-Altered vs. Human-Written LinkedIn Posts}},
  author = {Joos, Thomas Christian and Bouleimen, Azza and Hannák, Anikó and Pagan, Nicolò},
  year = {2026},
  doi = {10.48550/arXiv.2611.06660},
  eprint = {2611.06660},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.06660}
}

Reuse this entry