Under pressure: how time constraints, task complexity, and AI reliability shape human-AI interaction

This study investigates the impact of time pressure on human-AI collaboration in knowledge work environments. Drawing on dual-processing frameworks, specifically the heuristic-systematic model, we conducted an online experiment (n = 228) in which participants evaluated AI-generated responses across a series of tasks varying in complexity and correctness of AI responses. Participants subjected to a time-pressured treatment were compared to a control group with unlimited evaluation time. Our findings reveal that time pressure shifts cognitive processing from a systematic to a heuristic mode as expected, leading to diminished performance due to a reduced capacity to discriminate between correct and faulty AI responses. However, while both increased task complexity and AI advice faultiness independently impair human-AI-team performance, their interaction with time pressure suggests a more nuanced effect than predicted by traditional models. The results contribute to a refined understanding of cognitive dynamics in human-AI teaming and offer practical insights for designing AI systems and work environments that support effective decision-making under pressure.