Post AhvMBlwH5UIh0NErUe by arjen@idf.social
 (DIR) More posts by arjen@idf.social
 (DIR) Post #AhvMBlwH5UIh0NErUe by arjen@idf.social
       2024-05-15T17:44:50Z
       
       1 likes, 0 repeats
       
       Just believing that an AI is helping boosts your performancehttps://www.aalto.fi/en/news/just-believing-that-an-ai-is-helping-boosts-your-performanceResearchers discover an **AI placebo effect** where task performance improves when people believe an AI helps them. "The results also pose a significant challenge for research on HCI, since expectations would influence the outcome unless placebo control studies were used.‘These results suggest that many studies in the field may have been skewed in favor of AI systems,’ concludes Welsch."
       
 (DIR) Post #AhvMBmpZlsXJlt71gu by djoerd@idf.social
       2024-05-15T18:50:14Z
       
       0 likes, 0 repeats
       
       @arjen Carlos Castillo mentioned something similar in their #ECIR2024 keynote, where users reported they liked help of "AI" even if it  provided useless recommendations.
       
 (DIR) Post #AhvN3RypVfmek21tWC by djoerd@idf.social
       2024-05-15T18:59:55Z
       
       0 likes, 0 repeats
       
       @arjen Found it:  "users are only partially aware of the (lack of) accuracy of the decision support system."https://arxiv.org/abs/2203.15514
       
 (DIR) Post #Ahwk6YADT4AItNlKnw by arjen@idf.social
       2024-05-16T10:52:54Z
       
       0 likes, 0 repeats
       
       @djoerd yes, but the CHI paper shows that even when users only think they get AI support (a purely random signal) they still perform the task better.It's the "white coats in the factory" all over again; the Hawthorn effect (productivity may rise simply because of awareness of "being studied"); Kelloggs finding that wearing labcoats leads to better task performance; and Milgram's controversial Stanford experiments where subjects followed instructions blindly.Only, our white labcoats are AI!