Generative AI is transforming creative disciplines at an unprecedented rate, however there is a lack of studies investigating human-AI interaction in art and design education. To help close the research gap, this study takes a mixed-methods approach to evaluate the usability of text-to-image generators in design processes. The results from the System Usability Scale and semi-structured interviews indicate that while design students perceive the overall usability of generative AI positively, they rate the functionality, system feedback and efficiency of these tools less favorably. In-depth analyses of interview results suggest that limited control, limited interpretability and inconsistency of results are barriers for the usability of generative AI in design education.