Peer-reviewed research
Co-Writing with Opinionated Language Models Affects Users' Views
Maurice Jakesch et al., Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (ACM), 2023 · DOI 10.1145/3544548.3581196
Read the originalRead on October 16, 2026
- What it is
- An online experiment in which 1,506 participants wrote a short post with a language model writing assistant that argued one side of whether social media is good for society. The judges' ratings and a later attitude survey measured how the suggestions changed what people wrote and believed.
- How much weight it carries
- Peer-reviewed CHI paper. The task was a single short essay with one assistant configuration, and the attitude effects were measured right after writing.
Facts taken from this source
- The Jakesch et al. 2023 study was an online experiment with 1,506 participants who wrote a post about whether social media is good for society.
“In an online experiment, we asked participants (N=1,506) to write a post discussing whether social media is good for society.” (Abstract)
- Treatment participants in the Jakesch et al. study were shown suggested text generated by a large language model.
“Treatment group participants were shown suggested text generated by a large language model” (Introduction)
- Using the opinionated language model affected the opinions expressed in participants' writing, and it shifted their opinions in a later attitude survey.
“Using the opinionated language model affected the opinions expressed in participants” (Abstract)