From Paper to Card: Transforming Design Implications with Generative AI

TL;DR

Design implications in HCI are often overlooked by practitioners. The authors introduce design cards and an end-to-end AI-assisted system that generates these cards from scholarly papers using large language models and text-to-image models. Evaluations with designers (N=21) and paper authors (N=12) show AI-generated cards increase inspirability and generativity without sacrificing validity or originality, and authors view the tool as effective for communicating contributions. The work provides detailed guidance on card components, prompting strategies, and a pathway for scalable translation of research into practice, with noted opportunities for refinement and author collaboration.

Abstract

Communicating design implications is common within the HCI community when publishing academic papers, yet these papers are rarely read and used by designers. One solution is to use design cards as a form of translational resource that communicates valuable insights from papers in a more digestible and accessible format to assist in design processes. However, creating design cards can be time-consuming, and authors may lack the resources/know-how to produce cards. Through an iterative design process, we built a system that helps create design cards from academic papers using an LLM and text-to-image model. Our evaluation with designers (N=21) and authors of selected papers (N=12) revealed that designers perceived the design implications from our design cards as more inspiring and generative, compared to reading original paper texts, and the authors viewed our system as an effective way of communicating their design implications. We also propose future enhancements for AI-generated design cards.

Figures (6)

• Figure 1: A design card built on Baughan et al. (2020) baughan2020keep using our system
• Figure 2: An illustration of the Miro board participants used during the preliminary study. Each participant was asked to choose one among several options for each component, and drag & drop it to a card template to make a design card.
• Figure 3: Overview of the pipeline for generating design cards using generative AI models. In this case, we use GPT-3 as the LLM and DALLE-2 as the text-to-image model.
• Figure 4: Perceived qualities of design implication for each format using metrics from Sas et al. (2014) sas2014generating. Significance levels are based on the generalized linear mixed-effects modeling, and the bars indicate standard errors.
• Figure 5: Examples of misalignments surfaced in our survey