Collaborative virtual agents help human operators to perform tasks in real-time. For this collaboration to be effective, human operators must appropriately trust the agent(s) they are interacting with. Multiple factors influence trust, such as the context of interaction, prior experiences with automated systems and the quality of the help offered by agents in terms of its transparency and performance. Most of the literature on trust in automation identified the performance of the agent as a key factor influencing trust. However, other work has shown that the behavior of the agent, type of the agent’s errors, and predictability of the agent’s actions can influence the likelihood of the user’s reliance on the agent and efficiency of tasks completion. Our work focuses on how agents’ predictability affects cognitive load, performance and users’ trust in a real-time human-agent collaborative task. We used an interactive aiming task where participants had to collaborate with different agents that varied in terms of their predictability and performance. This setup uses behavioral information (such as task performance and reliance on the agent) as well as standardized survey instruments to estimate participants’ reported trust in the agent, cognitive load and perception of task difficulty. Thirty participants took part in our lab-based study. Our results showed that agents with more predictable behaviors have a more positive impact on task performance, reliance and trust while reducing cognitive workload. In addition, we investigated the human-agent trust relationship by creating models that could predict participants’ trust ratings using interaction data. We found that we could reliably estimate participants’ reported trust in the agents using information related to performance, task difficulty and reliance. This study provides insights on behavioral factors that are the most meaningful to anticipate complacent or distrusting attitudes toward automation. With this work, we seek to pave the way for the development of trust-aware agents capable of responding more appropriately to users by being able to monitor components of the human-agent relationships that are the most salient for trust calibration.

No abstract

Previous research indicates that synthetic speech can be as persuasive as human speech. However, there is a lack of empirical validation on interactive goal-oriented tasks. In our two-stage study (online listening test and lab evaluation), we compared participants' perception of the persuasiveness of synthetic voices created from speech in a debating style vs. speech from audio-books. Participants interacted with our Conversational Agent (CA) to complete 4 flight-booking tasks and were asked to evaluate the voice, message and perceived personal qualities. We found that participants who interacted with the CA using the voice created from debating style speech rated it as significantly more truthful and more involved than the CA using the audio-book-based voice. However, there was no difference in how frequently each group followed the CA's recommendations. We hope our investigation will provoke discussion about the impact of different synthetic voices on users' perceptions of CAs in goal-oriented tasks.

Trust is a prerequisite for effective human-agent collaboration. While past work has studied how trust relates to an agent's reliability, it has been mainly carried out in turn based scenarios, rather than during real-time ones. Previous research identified the performance of an agent as a key factor influencing trust. In this work, we posit that an agent's predictability also plays an important role in the trust relationship, which may be observed based on users' interactions. We designed a 2x2 within-groups experiment with two baseline conditions: (1) no agent (users' individual performance), and (2) near-flawless agent (upper bound). Participants took part in an interactive aiming task where they had to collaborate with different agents that varied in terms of their predictability, and were controlled in terms of their performance. Our results show that agents whose behaviours are easier to predict have a more positive impact on task performance, reliance and trust while reducing cognitive workload. In addition, we modelled the human-agent trust relationship and demonstrated that it is possible to reliably predict users' trust ratings using real-time interaction data. This work seeks to pave the way for the development of trust-aware agents capable of adapting and responding more appropriately to users. CCS CONCEPTS • Human-centered computing → Empirical studies in HCI; Collaborative interaction; User studies; Laboratory experiments.

No abstract