π€ AI Summary
This study addresses the limitation of existing AI alignment research, which predominantly relies on static pretraining paradigms while neglecting the dynamic evolution of humanβAI interactions within long-term post-deployment social contexts. To bridge this gap, this work reconceptualizes human-agent alignment as a continuous interaction design problem, proposing analytical methods for interactive system design and multi-turn feedback mechanisms to establish a dynamic alignment framework grounded in real-time interaction and trust evolution. The primary contribution is the development of SPEAR, a five-pillar framework encompassing Specification, Process, Evaluation, Adaptation, and Recalibration. By transcending the constraints of traditional static alignment approaches, this framework provides systematic theoretical guidance for humanβAI collaboration in complex scenarios.
π Abstract
Recent AI alignment work often frames alignment as a pre-deployment optimization problem: collect human feedback, learn preferences or principles, finetune the model, and deploy an aligned system. This framing has produced major progress, but it under-specifies what happens once AI systems act as agents on users' behalf in situated, long-term, and social contexts. This position paper reframes human-agent alignment as an ongoing interaction design problem. We propose SPEAR, five pillars of interactive alignment: Specification (how people express intent and establish shared understanding), Process (how agents decide when to act, ask, defer, or pause), Evaluation (how people judge whether agents succeeded), Adaptation (how agents adapt to users over repeated use), and Recalibration (how people adapt their trust, expectations, and behavior in response to agents).