Conversational User Interfaces: Explanations and Interactivity Positively Influence Advice Taking from Generative Artificial Intelligence DOI Open Access
Tobias R. Rebholz, Alena Koop, Mandy Hütter

и другие.

Опубликована: Март 28, 2024

Generative artificial intelligence (GenAI) has surged in popularity with the implementation of conversational user interfaces. The possibility to engage natural conversations GenAI thus constitutes a promising countermeasure algorithm aversion, or general preference for humans over algorithms, which hinders effective utilization often superior algorithmic output. In this study, we experimentally test influence explanations and interactivity on advice taking from GenAIs. judge-advisor system, 472 participants (313 female, 154 male, 5 diverse; median age 23 years) solved series ten estimation tasks access pre-generated output ChatGPT. control condition, only numerical was provided as advice. Participants treatment conditions were additionally with, could request, detailed explanation rationale underlying ChatGPT’s judgments. weight positively influenced by both opportunity interact receipt an explanation. Moreover, actively requesting significantly enhanced positive effect compared trials forgone. However, there no evidence that weighting whether requested. inherent explanatory capabilities interactively it independently increase users’ taking. This finding underscores potential interfaces large language models such ChatGPT improve individuals’ augmented judgment decision-making, but also poses threat human autonomy interactions systems.

Язык: Английский

Conversational User Interfaces: Explanations and Interactivity Positively Influence Advice Taking from Generative Artificial Intelligence DOI Open Access
Tobias R. Rebholz, Alena Koop, Mandy Hütter

и другие.

Опубликована: Март 28, 2024

Generative artificial intelligence (GenAI) has surged in popularity with the implementation of conversational user interfaces. The possibility to engage natural conversations GenAI thus constitutes a promising countermeasure algorithm aversion, or general preference for humans over algorithms, which hinders effective utilization often superior algorithmic output. In this study, we experimentally test influence explanations and interactivity on advice taking from GenAIs. judge-advisor system, 472 participants (313 female, 154 male, 5 diverse; median age 23 years) solved series ten estimation tasks access pre-generated output ChatGPT. control condition, only numerical was provided as advice. Participants treatment conditions were additionally with, could request, detailed explanation rationale underlying ChatGPT’s judgments. weight positively influenced by both opportunity interact receipt an explanation. Moreover, actively requesting significantly enhanced positive effect compared trials forgone. However, there no evidence that weighting whether requested. inherent explanatory capabilities interactively it independently increase users’ taking. This finding underscores potential interfaces large language models such ChatGPT improve individuals’ augmented judgment decision-making, but also poses threat human autonomy interactions systems.

Язык: Английский

Процитировано

1