In the context of design education, this study explores how a non-directive social robot (Furhat) can support early-stage idea generation. The robot was intentionally designed not to offer suggestions or solutions, distinguishing it from typical generative AI tools. Instead, it engaged students using metaphors, open-ended questions, and active listening. The aim was to examine how participants responded to a system that deliberately resisted the generative behavior commonly associated with tools like ChatGPT. Many participants appreciated the robot’s reflective role, noting that it helped them clarify and structure ideas without imposing direction. However, as the design process progressed, some expressed frustration and a desire for more proactive input—ranging from creative suggestions and emotional feedback to auditory elements, such as singing or whistling a melody. This mismatch between the robot’s expressive embodiment and its intentionally limited functionality led to moments of tension. Rather than viewing this as a shortcoming, we frame it as a deliberate design strategy: by withholding generative output, the robot fosters independent ideation. These findings contribute to broader discussions in human-robot interaction, particularly concerning creativity, user expectations, and the design of non-directive AI systems.