This article discusses a study comparing users’ choices between performing tasks themselves and delegating them to an AI agent in an environment where large language models (LLMs) and graphical interfaces are used together.