Better Slow than Sorry: Introducing Positive Friction for Reliable Dialogue Systems
🤖 AI Summary
Contemporary AI dialogue systems overemphasize response speed and superficial fluency, fostering user overtrust, obscuring implicit assumptions, and compromising task accuracy and reliability. To address this, we propose the “constructive friction” paradigm: deliberately introducing interventions—such as targeted queries, explicit assumption articulation, or strategic response pauses—at critical decision points to prompt user reflection and goal recalibration. We present the first systematic ontology of constructive friction, translating cognitive science’s deliberative pacing theory into a modelable, triggerable dialogue regulation framework. Our approach integrates expert-annotated multilingual corpora, state-aware triggering policies, and large language model–driven assumption inference and reflective prompting. Experiments across multiple benchmarks and simulated tasks demonstrate significant improvements: +12.7% in task success rate, enhanced user goal alignment and belief modeling accuracy, and increased interpretability and accountability of AI decisions.
Application Category
📝 Abstract
While theories of discourse and cognitive science have long recognized the value of unhurried pacing, recent dialogue research tends to minimize friction in conversational systems. Yet, frictionless dialogue risks fostering uncritical reliance on AI outputs, which can obscure implicit assumptions and lead to unintended consequences. To meet this challenge, we propose integrating positive friction into conversational AI, which promotes user reflection on goals, critical thinking on system response, and subsequent re-conditioning of AI systems. We hypothesize systems can improve goal alignment, modeling of user mental states, and task success by deliberately slowing down conversations in strategic moments to ask questions, reveal assumptions, or pause. We present an ontology of positive friction and collect expert human annotations on multi-domain and embodied goal-oriented corpora. Experiments on these corpora, along with simulated interactions using state-of-the-art systems, suggest incorporating friction not only fosters accountable decision-making, but also enhances machine understanding of user beliefs and goals, and increases task success rates.
Problem
Research questions and friction points this paper is trying to address.
AI Dialogue Systems
User Trust
Accuracy and Reliability
Innovation
Methods, ideas, or system contributions that make the work stand out.
Positive Friction
AI Dialogue Systems
Enhanced Reliability
🔎 Similar Papers
No similar papers found.
