What is it about?
Interactions of users and AI systems are becoming nearly as complex as human-to-human interactions; in AI Experience Design (AIX), both the human and the AI can both dynamically change their behaviors. For the designer, prototyping AI interfaces is much more complex because divergent AI behaviors must be considered for diverse end-users, user histories, and use contexts. The Model-Informed Prototyping process links simultaneous exploration of AI behaviors and alternative design features. Our ProtoAI tool directly incorporates AI model outputs into a prototype interface design to test AI behavior outcomes across differing end-user data contexts,. This allows AIX designers to anticipate and address potential AI performance, errors, breakdowns, and feedback for varied users within a rapid, iterative prototyping process.
Featured Image
Photo by Vitolda Klein on Unsplash
Why is it important?
AI systems produce dynamic changes in behavior due to machine learning from user data. Designing AI user experiences (AIX) requires anticipating and addressing potential changes in AI behaviors across diverse end-users and use contexts. Our prototyping process and tool support simultaneous exploration of AI behavior and interface design choices across diverse users and contexts, providing rapid, iterative prototyping of dynamic AI experiences.
Perspectives
The challenges in AI experience design (AIX) are game-changing because like users, AI systems change with use. This model and ProtoAI tool provide a tractable method for rapid, iterative prototyping when both humans and AI behaviors are dynamic.
Colleen Seifert
University of Michigan, Ann Arbor
Read the Original
This page is a summary of: ProtoAI: Model-Informed Prototyping for AI-Powered Interfaces, April 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3397481.3450640.
You can read the full text:
Contributors
The following have contributed to this page
