From Pattern Libraries to Practice: Context-Aware Selection of AI Interface Patterns

AI interface pattern libraries provide valuable catalogs, but practitioners lack decision support for selecting patterns that fit specific contexts. We introduce a context-aware selection framework that translates situational factors (user, task, interface, regulatory context) into priorities over four design dimensions: trust calibration, user control, transparency, and personalization, linking these to concrete pattern choices. We apply the framework to AI-assisted form authoring, where high subjectivity and data-protection constraints prioritize user control, motivating per-field, on-demand suggestions over system-initiated automation. A preliminary mixed-method pilot (N = 6) provides converging signals: four participants preferred on-demand suggestions and described an authoring stance (“choosing a direction”) rather than editing (“checking for errors”). These findings motivate a larger cross-domain study to test whether coarse context assessments predict calibrated reliance behavior.