Enabling Morally Sensitive Robotic Clarification Requests

Abstract

The design of current natural language-oriented robot architectures enables certain architectural components to circumvent moral reasoning capabilities. One example of this is reflexive generation of clarification requests as soon as referential ambiguity is detected in a human utterance. As shown in previous research, this can lead robots to (1) miscommunicate their moral dispositions and (2) weaken human perception or application of moral norms within their current context. We present a solution to these problems by performing moral reasoning on each potential disambiguation of an ambiguous human utterance and responding accordingly, rather than immediately and naively requesting clarification. We implement our solution in the Distributed Integrated Cognition Affect and Reflection robot architecture, which, to our knowledge, is the only current robot architecture with both moral reasoning and clarification request generation capabilities. We then evaluate our method with a human subjects experiment, the results of which indicate that our approach successfully ameliorates the two identified concerns.

Do as I Demand, Not as I Say: A Dataset for Developing a Reflective Life-Support Robot

“…As shown in FIGURE 1, we presented videos of the robot performing defined actions to facilitate worker understanding of the robot actions. We used crowdsourcing 3 to collect a Japanese corpus based on the robot-action categories defined in TABLE 2. Note that our data would inevitably be biased by the Japanese cultural background relevant to watching other people's surrounding situations.…”

“…Interactive robots that cooperate with humans must take appropriate actions in response to user requests. Existing human-computer interaction systems, which cooperate with human users and include interactive robots, assume that users explicitly verbalize their requests and that systems can confirm ambiguous requests with the users [1], [2], [3]. However, user requests often have information gaps in their actual demands because it is challenging to verbalize requirements without over-or underestimating them [4], [5].…”

Robots and Robotics in Nursing

“…Approaches to deepening understanding by having a robot repeatedly ask questions in response to unintended utterances [ 58 ] and methods to perform moral reasoning to resolve the ambiguity without asking questions [ 59 ] have also been studied. However, it is difficult for robots to make judgments that humans can make from context.…”

Exploring the Naturalness of Cognitive Status-Informed Referring Form Selection Models

“…While generating effective definite descriptions is a critical task, a speaker solely relying on this referring form would be an inefficient, unnatural, and annoying speaker. This discrepancy is critical not just for the psycholinguistics community, who seek to understand the cognitive dynamics of language production, but also for the Artificial Intelligence and Human-Robot Interaction communities, who seek to enable efficient, natural, and humanlike communication in task-based, situated domains (Tellex et al, 2013;Jackson and Williams, 2022;Cakmak and Thomaz, 2012;Williams et al, 2015;Gervits et al, 2021). As such, we argue that more attention to the problem of Referring Form Selection is needed across multiple areas of cognitive science.…”