Countering climate misinformation with large language models: Evidence from ChatGPT
- 1Department of Communication Studies, School of Communication, Hong Kong Baptist University, Hong Kong, China
Abstract
Public support for climate action hinges on the integrity of information environments. In an online experiment, U.S. adults used ChatGPT to evaluate climate-related claims. We first conducted computational text analysis of GPT conversation logs, assessing valence, formality, bias, recency, authority cues, and semantic richness of ChatGPT’s responses using language-model based classifiers and link-level metadata. Concurrently, we measured credibility judgments for both the claims and ChatGPT’s responses through participant self-reports. Credibility perceptions were driven primarily by individual differences; greater acceptance of the scientific consensus, attention to climate change, prior familiarity with ChatGPT, and younger age predicted higher perceived credibility. Once these factors were accounted for, authority cues were associated with slightly less extreme climate attitudes, while positive valence and semantic richness correlated positively with perceived credibility. These results support transparent, unbiased, audience-tailored engagement as a pathway to strengthen responsible climate communication and help bridge the gap between scientific consensus and public understanding.
Related
- Evidence for climate-mediated range expansion of hybrid wood ants
- Climate change adaptation barriers among beekeepers across agroecological zones in Gedeo, Ethiopia: evidence from problem confrontation index and multinomial regression
- What Transfers Across Brains Is Not What Stays Within Individuals: Evidence from Bilingual Language States
- Linking nutritional quality and climate impact of dairy and plant-based milk and yogurt alternatives: Evidence from three countries.
- Grounding-Specific Effects of Valence in Word Recognition: Evidence from Pain-Related Language.
BibTeX
@article{tsang2026countering,
title = {{Countering climate misinformation with large language models: Evidence from ChatGPT}},
author = {Stephanie Jean Tsang and Dandan Wang},
journal = {PLOS Climate},
year = {2026},
month = sep,
doi = {10.1371/journal.pclm.0000930},
url = {https://journals.plos.org/climate/article?id=10.1371/journal.pclm.0000930},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Countering climate misinformation with large language models: Evidence from ChatGPT
AU - Stephanie Jean Tsang
AU - Dandan Wang
PY - 2026
DA - 2026/09/02
JO - PLOS Climate
DO - 10.1371/journal.pclm.0000930
UR - https://journals.plos.org/climate/article?id=10.1371/journal.pclm.0000930
AB - Public support for climate action hinges on the integrity of information environments. In an online experiment, U.S. adults used ChatGPT to evaluate climate-related claims. We first conducted computational text analysis of GPT conversation logs, assessing valence, formality, bias, recency, authority cues, and semantic richness of ChatGPT’s responses using language-model based classifiers and link-level metadata. Concurrently, we measured credibility judgments for both the claims and ChatGPT’s responses through participant self-reports. Credibility perceptions were driven primarily by individual differences; greater acceptance of the scientific consensus, attention to climate change, prior familiarity with ChatGPT, and younger age predicted higher perceived credibility. Once these factors were accounted for, authority cues were associated with slightly less extreme climate attitudes, while positive valence and semantic richness correlated positively with perceived credibility. These results support transparent, unbiased, audience-tailored engagement as a pathway to strengthen responsible climate communication and help bridge the gap between scientific consensus and public understanding.
SN - 2767-3200
ER -