This qualitative case study explored graduate students' views (n=6) on AI-supported applications and an AI-enabled blockbased coding tool (PictoBlox) in science education. Data were gathered over a 39-hour implementation via a semi-structured interview form and screen captures from the activities, and analyzed with content analysis. Participants perceived AI tools as time-saving and pedagogically enriching, while emphasizing ethics and data security. PictoBlox's AI add-ons (e.g., natural language processing, image processing, machine learning) were seen to support concretization, visualization, and interactive content creation. Reported challenges concerned activity design and block creation skills. We discuss implications for teacher education, including targeted training on AI-supported lesson planning and assessment design, and guidance on ethical/data-protection practices. Limitations (convenience sampling, small n, self-report) constrain generalizability. Future research should replicate with larger, diverse cohorts and triangulate with classroom observations.