CogXR: An AI/XR Powered Graphic Tool Enhancing Human Rationality

CogXR is an AI-powered learning companion in academic and professional contexts designed to generate interactive visualized reasoning graphics to help people navigate the complexities of academic concepts easily. For the current prototyping phase, we are using the Quivr API as the fundamental LLM for input processing and Blender for content-based interactive graphics production and XR implementation. The core research phase to be present on the HCII 2024 International Conference on Augmented Cognition is the use test of improvements in reading cognition with Cog-XR. The fundamental goal here is to validate the level of improvements in cognitive ease in the reading process with the assistance of AI-generated content-based graphics. We plan to conduct the test in academic reading contexts involving interactive sessions with the user groups. We will conduct a comparative user test with two sample groups. Each group will complete a short reading task, take a quiz based on a specific article, and participate in a brief user experience interview. One group will work with Cog-XR visualizations to aid their reading, while the other group will use text abstracts generated by currently trending chat-box PDF readers as assistance. The potential outcomes of the research phase will be the quantitative analysis of the improvement in cognitive ease of the reading process with assistance from AI-generated graphics. Such outcomes could potentially serve as validation for utilizing LLMs to produce visual and graphic interventions to improve human rationality and enhance cognitive ease in professional reading contexts.