Users value conditions set around AI agent use more than outcomes
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw
Human-Computer InteractionArtificial Intelligence
Summary
People often let AI agents do tasks for them, but usually we only check if the task was finished. The authors studied lots of user posts about an AI agent called OpenClaw to see what values users cared about. They found that users cared more about how they set up and monitored the AI, like costs and fairness, rather than just what the AI delivered. This means creating AI tools also means paying attention to how people control and oversee them.
What this means in practice
- •For ai product designers: Design AI systems that support user control over conditions like costs, access, and supervision to better align with human values in delegation.
- •For customer support managers: Develop monitoring protocols that help users supervise AI agents effectively, addressing value concerns across autonomy and fairness.
Authors
Renkai Ma, Ruyuan Wan, Xuan Lu, Fan Yang, Chen Chen, Lingyao Li
Abstract
Users increasingly delegate work to autonomous AI agents, yet evaluations typically measure task completion rather than the values users prioritize. Using Value Sensitive Design, we analyzed, with LLM assistance, 73,093 first-person Reddit posts about using OpenClaw, each for its human value, agent aspect, value fulfillment, and user outcome. The 21 values form six value groups, including Autonomous, Dependable, and Affordable Operation, Bounded Reach, Reviewability, and Equitable Access. Relative to each aspect's corpus share, values clustered not at the agent's outputs but at the operating conditions users set around a run. Values were usually met where users described what the agent delivered, in five of six groups, and mostly unmet where users described supervising it, in all six groups. We conceptualize this pattern as value-sensitive delegation. Supporting human values requires attention not only to what an agent accomplishes, but to the conditions users set around delegation, including cost, access, and oversight.