Abstract:
This study examines the 'Phantom AI effect', a cognitive bias wherein members of remote creative teams mistakenly attribute human-generated ideas or feedback to Artificial Intelligence (AI) in the absence of any AI use. As generative AI tools become prevalent in South African workplaces, this misattribution poses unexamined psychosocial risks. Using a sample of 412 South African creative professionals and Structural Equation Modelling (SEM) with AMOS, the study finds that Phantom AI attribution significantly reduces team trust (β = -0.52, p < 0.001), diminishes creative self-efficacy (β = -0.48, p < 0.001), and negatively affects project outcomes (β = -0.41, p < 0.001). Automation bias and limited AI literacy are identified as primary antecedents. The study introduces the Phantom AI effect as a distinct phenomenon, contributes to organisational psychology and human-AI interaction literature, and offers practical recommendations for managing remote creative teams.
