This work-in-progress study examines how undergraduate design students interact with AI-powered large language models (LLMs) such as ChatGPT and Gemini during the design process. Using a mixed-methods approach grounded in Technology Acceptance Model, Human Centred AI framework, and Emotional Design theory, we combined surveys (n = 51) with semi-structured interviews. Findings show that while students find LLMs useful and easy to use, their experiences involve ethical concerns, emotional motivations, and creative tensions. Students used LLMs not just for ideation, but also as collaborators and cognitive supports. Some expressed discomfort with originality, memory limitations, or ideological values of LLM providers. These insights highlight the need for more adaptive, emotionally aware AI tools in education. Data collection is ongoing, with additional interviews planned.