From Paper to Card: Transforming Design Implications with Generative AI Donghoon Shin, Lucy Lu Wang, and Gary Hsieh present a system that uses generative AI models to create design cards from academic papers, making design implications more accessible and actionable for designers. The system leverages large language models (LLMs) and text-to-image models to generate design cards that include title, description, image, paper summary, evidence, and citation. The design cards were evaluated by 21 designers and 12 paper authors, who found that the design cards were more inspiring and generative compared to reading the original paper texts. Authors viewed the system as an effective way to communicate their design implications. The study highlights the potential of AI-generated design cards to enhance the delivery of design insights and suggests future improvements, such as refining the pipeline for extracting evidence and enhancing the use of images. The system is implemented as a web interface that allows users to upload a paper PDF and generate a design card. The system uses Grobid and BeautifulSoup to parse the paper and extract components. The results show that design cards were perceived as more valid, generalizable, and original compared to raw text. The study contributes to the field of human-centered computing by providing a scalable and efficient way to translate academic findings into a prescriptive format such as design cards.From Paper to Card: Transforming Design Implications with Generative AI Donghoon Shin, Lucy Lu Wang, and Gary Hsieh present a system that uses generative AI models to create design cards from academic papers, making design implications more accessible and actionable for designers. The system leverages large language models (LLMs) and text-to-image models to generate design cards that include title, description, image, paper summary, evidence, and citation. The design cards were evaluated by 21 designers and 12 paper authors, who found that the design cards were more inspiring and generative compared to reading the original paper texts. Authors viewed the system as an effective way to communicate their design implications. The study highlights the potential of AI-generated design cards to enhance the delivery of design insights and suggests future improvements, such as refining the pipeline for extracting evidence and enhancing the use of images. The system is implemented as a web interface that allows users to upload a paper PDF and generate a design card. The system uses Grobid and BeautifulSoup to parse the paper and extract components. The results show that design cards were perceived as more valid, generalizable, and original compared to raw text. The study contributes to the field of human-centered computing by providing a scalable and efficient way to translate academic findings into a prescriptive format such as design cards.