This project explores the challenges and design requirements involved in developing safe and user-centered experiences for autonomous AI agents. Unlike traditional AI that simply responds to user prompts, agentic AI (e.g. Anthropic’s Claude) can independently interact with computers and carry out complex workflows without continuous user input. While these advancements offer increased productivity and convenience, they also introduce new risks, such as loss of user control, transparency concerns, and potential ethical issues. This research team will conduct literature reviews to establish a clear definition of “AI agents,” given the variety of interpretations, then conduct an in-depth UX analysis, comparative analysis, and data analysis to identify challenges and key moments where users may feel uncertain or lose their sense of control over the agent’s autonomous actions. The expected outcome is a research paper with design recommendations for future UX/UI designers working on agentic AI products.
