Develop an interface for model-informed prototyping of HAI interactions

More Info


As AI technology continues to advance, there's a growing need to integrate it into UX design. However, AI's unique characteristics do not seamlessly align with current design tools and mastering the technical aspects for designers is a significant challenge. The project goal is to develop a tool based on a developed semi-formal representation for Human-AI (HAI) interactions, which uses a set of communicative acts1 to specify the communicated information between users and AI models as exchanges of messages.

The project followed an iterative prototyping method across 4 phases. The Pre-Phase aimed at testing communicative acts with design students using a use case ("CV-Screening") and paper materials. At the same time, it also expected to get insights on the data structure and develop the specific design considerations based on those for the Model-Informed Prototyping (MIP),

Phase 1 explored effective workflows of the digital prototype to present communicative acts by following the design considerations from the Pre-Phase and using the low-fidelity digital prototype in Figma. The use case in this phase was the same as that in the Pre-Phase.

In Phase 2, based on insights from the last two phases, there was a high-fidelity prototype in Figma which was inspired by the user journey map. It was used to assess if the design output achieved the design goals and considerations, and it helped the final test materials work better.

The Final Phase utilizes the refined digital prototype for the final test which had the same goals as that in Phase 2, providing important insights for future development.

The final output of the project is a partial prototype of a digital tool designed to facilitate the early stages of human-AI interaction design. Grounded in the principles of communicative acts and human-centered design, this tool assists designers during the Ideation stage of Design process. It achieves this by visualizing the roles, data, and information involved in the process of information exchange during Human-AI Interactions. The goal is to enhance efficiency and ease in designing these interactions.