Designing legal interfaces: Human-machine agency and the practice of AI in U.S. law

In U.S. legal systems, agency has long been anchored in intent, responsibility, and accountability. As AI becomes embedded in legal workflows, these boundaries increasingly blur. Although AI systems lack consciousness, they shape legal reasoning and influence judicial outcomes, functioning as consequential non-human agents. In this article we examine how such systems are interpreted, trusted, and held to account within culturally symbolic and institutionally rigid legal environments. We investigate how legal professionals ascribe agency to AI (granting it influence, credibility, and, at times, blame) and argue that AI’s role emerges not only from its outputs but from the human responses that give those outputs meaning and authority. For UX researchers, this raises a critical design challenge: when we broker agency to non-human actors through interfaces, language models, and interaction patterns, what are the implications for justice, truth, and the design of systems that shape human outcomes?