Moritz Joerling, Professor at emlyon business school and his co-authors, have published an article in Communications Psychology, a new journal from Springer Nature: People devalue generative AI's competence but not its advice in addressing societal and personal challenges.

According to this new open access research, individuals view ChatGPT 3.5 as a source of advice on how to tackle societal and personal challenges — such as climate change, pandemic preparedness, or improving individual health — as less competent than human advisors. However, people appear to still consider AI-generated advice as valuable and are equally likely to share it with friends and family.

Moritz Joerling and his co-authors conducted a series of experiments to investigate how individuals' perceptions of ChatGPT advice varied depending on whether they knew the advice was AI-generated, the subject matter being advised on, and their previous interactions with AI-generated advice. The authors found that participants consider AI to be less competent at providing advice than humans when they know the advice came from an AI; however, the actual advice provided was still viewed as useful and likely to be shared further. Furthermore, when participants had a positive experience of AI-generated content, they were more likely to select ChatGPT advice.

Together, these findings suggest that AI-generated advice could appear helpful to human decision makers when it comes to addressing societal and personal challenges.

Source: Böhm, R., Jörling, M., Reiter, L. et al. People devalue generative AI's competence but not its advice in addressing societal and personal challenges. Commun Psychol 1, 32 (2023). https://doi.org/10.1038/s44271-023-00032-x