Triadic Human–AI Collaboration in Technical Implementation for Creative Design: Cognitive Load, Team Workload, and Performance Analysis

Generative AI is increasingly embedded in design workflows, yet little is known about how novice designers collaborate with AI tools and human instructors. This study examined how novice design teams navigate technical implementation challenges through natural collaboration with both AI tools and instructors. Forty students in ten teams joined a four-week workshop to develop interactive installation demos. Quantitative results showed that instructor involvement increased students’ germane cognitive load, while intrinsic and extraneous cognitive load and team workload did not change significantly. Qualitative findings identified three collaboration models: co creation, task distribution, and responsibility-centered, associated with different performance levels. High performing teams and technically stronger students used AI mainly for technical support and instructors for design thinking. In contrast, medium and low performing teams and less technically skilled students relied more on both AI and instructors for problem solving and emotional support, highlighting the irreplaceable role of human instruction.