The Goldilocks Time Window for Proactive Interventions in Wearable AI Systems

As AI systems become increasingly integrated into our daily lives and into wearable form factors, there's a fundamental tension between their potential to proactively assist us and the risk of creating intrusive, dependency-forming experiences. This work proposes the concept of a Goldilocks Time Window -- a contextually adaptive time window for proactive AI systems to deliver effective interventions. We discuss the critical factors that determine the time window, and the need of a framework for designing and evaluating proactive AI systems that can navigate this tension successfully.