Generative AI usage and improvisation capability: The mediating role of technological affordances and the moderating role of task uncertainty

Generative artificial intelligence (AI) is increasingly used in knowledge-intensive work, yet its implications for employees’ improvisational capabilities—critical for adaptive and innovative outcomes—remain insufficiently understood. This study examines how three distinct patterns of generative AI use (task-oriented, creativity-oriented, and collaboration-oriented) relate to individual improvisation, mediated through key technological affordances (rapid ideation, content personalization, and collaborative support) and moderated by task uncertainty. Using a three-wave, time-lagged survey of 500 knowledge workers in China, we find that all three types of AI use are positively associated with improvisation. Technological affordances partially mediate these relationships, highlighting the mechanisms through which AI enables spontaneous problem-solving and innovation. Furthermore, task uncertainty strengthens the AI use–improvisation link, with effects substantially stronger in highly uncertain task environments. Our model explains 47% of the variance in improvisation capability. These findings contribute to research on technology-enabled adaptation and offer practical insights primarily for organizations in technology-driven and consulting-oriented knowledge-intensive sectors aiming to leverage generative AI for enhancing adaptive performance, with cautious extension to similar high-AI-adoption contexts.