Human heuristics for AI-generated language are flawed
By Maurice Jakesch, Jeffrey T. Hancock, Mor Naaman. Proceedings of the National Academy of Sciences of the United States of America, 2022
366 citations recorded by the source corpus.
Significance Human communication is now rife with language generated by AI. Every day, across the web, chat, email, and social media, AI systems produce billions of messages that could be perceived as created by humans. In this work, we analyze human judgments of self-presentations written by humans and generated by AI systems. We find that people cannot detect AI-generated self-presentations as their judgment is misguided by intuitive but flawed heuristics for AI-generated language. We demonstrate that AI systems can exploit these heuristics to produce text perceived as “more human than human.” Our results raise the question of how humanity will adapt to AI-generated text, illustrating the need to reorient the development of AI language systems to ensure that they support rather than undermine human cognition.
Subfields: Artificial Intelligence, Social Psychology, Sociology and Political Science
View Paper: https://www.semanticscholar.org/paper/98962528d113b47fd72ebe7ff205fb6a14d0d9e3
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Sociology and Political Science
Citations in the corpus
- Generative AI (2023, neutral): “, fake news, propaganda) that is hard to detect by humans (Jakesch et al., 2023; Kreps et al., 2022).”
- Generative artificial intelligence (2023, background): “The promotion of explainable and responsible AI becomes paramount once more (Meske et al., 2022), especially in a technological state where distinguishing between generated and real content by GAI models is becoming increasingly more challenging (Jakesch et al., 2023b; Lehmann & Buschek, 2020).”