🤖 AI Summary
This work addresses the challenge of evaluating nonverbal interaction perception and behavioral credibility of embodied AI agents under human-induced stress. We propose “React to This” (RTT), a novel Turing test framework that shifts focus from textual comprehension to *reactivity*—specifically, the agent’s capacity to generate contextually appropriate, temporally grounded, and posture-aware nonverbal behaviors. Grounded in embodied cognition theory, RTT formalizes a perception–action closed-loop model and validates it through controlled human-subject experiments with expert annotators. Methodologically, RTT introduces a scalable, behaviorally grounded benchmark for assessing embodied AI trustworthiness beyond linguistic capability. Our contribution extends the theoretical scope of the Turing test while providing an operationally feasible, extensible evaluation protocol. Preliminary experiments demonstrate RTT’s sensitivity in exposing critical limitations—such as delayed response timing, poor situational adaptation, and inconsistent motor coordination—in current embodied agents, thereby identifying concrete avenues for improvement.
📝 Abstract
We propose an approach to test embodied AI agents for interaction awareness and believability, particularly in scenarios where humans push them to their limits. Turing introduced the Imitation Game as a way to explore the question: "Can machines think?" The Total Turing Test later expanded this concept beyond purely verbal communication, incorporating perceptual and physical interaction. Building on this, we propose a new guiding question: "Can machines react?" and introduce the React to This (RTT) test for nonverbal behaviors, presenting results from an initial experiment.