Decision-Making in the Era of AI Support-How Decision Environment and Individual Decision Preferences Affect Advice-Taking in Forecasts

Decisionmakers often receive advice from different forecast sources (e.g., human or machine), which they can integrate into their own assessments. Previous studies led to different results regarding the use of forecasts depending on their source. With the ongoing digitalization, it can be assumed, first, that forecasts will be used more frequently in general and, second, that forecasts will increasingly be created digitally or in an automated manner. To investigate these factors in more detail, a vignette experiment was conducted with over 600 participants. In particular, we focused on the decision preference of the forecast users to take a deeper look at the fit of a forecast source, decision preference, and advice-taking. The results show that both a longer time horizon as well as greater trust in advice increase the acceptance of advice. With regard to the forecast source, we find that in a long-term scenario the advice by artificial intelligence is preferred over a human advice. Concerning the relationship of trust and decision preference on the acceptance of advice, our results show that the impact of the forecast source on trust in advice is moderated by trust in digital technologies and that the moderation by trust in digital technologies is itself moderated by decision preference, which highlights the relevance of a fit between the forecast source and an individual's decision preference.