Mobile applications have become increasingly personalized due to the ability for a phone to record information like location, time, weather, and other personal metrics. App designers are now able to leverage this data to create an enhanced user experience. Context-aware applications add value to their users by triggering relevant actions in the context of a specific human experience. While a machine will never understand a human experience, a human designer can translate these experiences into a machine-executable representation. To create an application like this, designers must translate human experiences into concept expressions—machine-executable logical representations built from context features.

Designers must consider a wide range of context features relevant to the human experience. If a concept expression is too narrow, designers risk delivering limited or no results for some users.

Designers can struggle with idea fixation through a creative brainstorming process and can limit their own ability to create fully encompassing concept expressions. This project presents a system that leverages large language models like Chat GPT to help designers break free of design fixation, and expand their existing ideas spaces to create richer concept expressions. The core contribution of this work is a workflow that uses AI as a collaborator to iteratively generate ideas and reflect throughout the conception of human experiences. LLMs can very easily take different approaches to the same experience, making them powerful brainstorming assistants. By directly attempting new approaches, the LLM effectively avoids design fixation; we take advantage of that in our prototype. The findings so far show that by integrating AI into the brainstorming process, the system successfully generates ideas that are outside the scope of a designer's immediate thoughts, and can help potential designers break outside of idea spaces they may be fixating on.