Defining system requirements in engineering design has always been challenging and complex. This research explores the potential for Large Language Models to support and enhance requirements develop- ment. To explore this potential, a mixed-methods approach is employed, combining quantitative surveys and qualitative interviews with industry professionals who manage requirements. The original data set consisted of human-created system requirements, which were compared to AI-generated requirements that were assessed for completeness using four criteria: specificity, functionality, target values, and ver- ifiable. The interviews provided valuable insights into current workflows and the common challenges faced in requirement definition, and also the potential benefits and limitations of AI solutions. The re- sults indicated that AI-generated requirements can help make the process more manageable and act as a collaborative partner for human engineers. Although AI may miss some important details, there is still significant potential for its improvement to create models capable of accurately defining requirements.