Reciprocal human-AI learning in crowdsourced design: algorithm aversion, tacit knowledge, and ethical transparency

The fast integration of generative artificial intelligence in digital design has altered the use of AI systems from an automated tool to a partner. With this change, there is limited empirical research that can explain how adaptive human-AI interaction can lead to a better creative outcome, especially in crowdsourced design settings. This paper explores how reciprocal human-AI learning (RHML) influences creative synergy and considers the mediating and moderating roles of ethics, transparency, and algorithm aversion, respectively. A 2 x 2 factorial controlled experiment was carried out with 320 subjects with design backgrounds who completed an AI-assisted creative task under a simulated crowdsourced design condition. Expert raters assessed creative, productive outputs, and validated questionnaires assessed psychological constructs. The analysis was done using the factorial ANOVA and moderated mediation analysis with PROCESS. The findings reveal that RHML has a great effect on enhancing creative synergy (F(1,316) = 28.74, p < .001). RHML also decreases aversion to algorithms, which partly mediates its beneficial influence on the results of creativity. Moreover, ethical disclosure puts a lot of force in the correlation between RHML and creative synergy (interaction F(1,316) = 12.63, p < .001). The results prove that successful human-AI cooperation in creative systems is based not only on technological ability but also on psychological acceptance and clear governance mechanisms. The research paper provides empirical support to the study of hybrid intelligence and has design implications in the use of AI-assisted creative platforms.