Conversational agents are increasingly used to substitute or augment human service employees. Due to their ability to use human‐like communicative behaviors, these agents are theorized to establish ...social connections with customers. However, the existing literature is ambiguous on how conversational agents should verbally communicate with customers and whether they should be adaptive to customers' verbal behavior. The current study aims to address these gaps, by focusing on the effects of the conversational agent's communication style on the perceived social connection. Two experiments were conducted in which a virtual agent in a hedonic service (Experiment 1) or a utilitarian service (Experiment 2) was manipulated to adapt either a static task‐ or a social‐oriented communication style or mimic the communication style used by the customer in the previous turn. Guided by marketing and human–computer interaction literature, measures for engagement and rapport were used to reveal customers' perceived social connection. Results show that for the hedonic service, rapport was significantly affected by the presence of social cues in the agent's communication style (whether statically or mimicked), while engagement was significantly affected by mimicry. For the utilitarian service, only social cues significantly affected rapport. These findings enrich the Computers As Social Actors paradigm and provide clear guidelines for practitioners.
Embodied conversational agents are often included in health behaviour change applications as intelligent virtual coaches. A major challenge in their development is tailoring coaching dialogues to ...user profiles. Agents should collect information about the user and consequently adapt the strategy that guides their interactions.
Previous research discovered relations between users’ motivation profiles and potential effective coaching strategies. In the current paper, we describe an experiment with multiple agents that tests if users with certain motivation profiles prefer certain (tailored) strategies.
Participants were classified into four motivation groups (Intrinsic Motivation, External Regulation, Dual Motivation, A-motivation), following their responses to a questionnaire on motivation towards healthy living. Then, two coaches suggested a positively and a negatively tailored strategy. Participants rated these and chose their favourite.
Results (N = 108) show that the Dual Motivation group appreciated their positively tailored strategy more than their negatively tailored strategy, while intrinsically motivated participants appreciated both strategies. Furthermore, agents' likeability does not seem to influence strategy appreciation, while there was an effect of participant's age and gender.
We conclude that coaching strategies for dialogues with agents can be tailored to personal motivation to live healthy. Future research should focus on performing a long-term study in a real-life setting.
•Tailoring of coaching strategies is explored in a multi-agent application.•Personal motivation to live healthy can be used to tailor coaching strategies.•People can be both intrinsically motivated and externally regulated.•Participants' age and gender influence the appreciation of coaching strategies.•Likeability of the presenting agent did not influence strategy appreciation.
Virtual Humans (VHs) emerged over 50 years ago and have since experienced notable advancements. Initially, developing and animating VHs posed significant challenges. However, modern technology, both ...commercially available and freely accessible, has democratized the creation and animation processes, making them more accessible to users, programmers, and designers. These advancements have led to the replication of authentic traits and behaviors of real actors in VHs, resulting in visually convincing and behaviorally lifelike characters. As a consequence, many research areas arise as functional VH technologies. This paper explored the evolution of four areas and emerging trends related to VHs while examining some of the implications and challenges posed by highly realistic characters within these domains.
Display omitted
•We present a detailed foundation and history of Virtual Humans.•We conducted a literature review on four relevant areas for Virtual Humans•Areas are: Perception; Facial Transferring; Conversational Agents; and Crowd Simulation.•We discuss the main challenges still present in the four mentioned areas.
ABSTRACT Recent technological advances make it possible to create automated virtual interviewers, called embodied conversational agents (ECAs). We study how an ECA compares to a human interviewer in ...three experiments. In experiment 1, we show that two theoretically motivated factors—making the ECA facially and vocally similar to the interviewee—result in the ECA performing similarly to or better than human interviewers for six antecedents of disclosure quality. In two additional experiments, we show that employees are, on average, 21 to 32 percent more likely to disclose violating internal controls to an ECA than to a human, even if the human interviewer has significant interviewing experience. These findings contribute to the ECA design literature by showing that similarity-enhancing features of ECAs increase the antecedents of disclosure. The findings also contribute to the accounting literature by demonstrating that ECA technology can increase the scope of interviewing in accounting without reducing interview quality.
Chatbots are intelligent conversational computer systems designed to mimic human conversation to enable automated online guidance and support. The increased benefits of chatbots led to their wide ...adoption by many industries in order to provide virtual assistance to customers. Chatbots utilise methods and algorithms from two Artificial Intelligence domains: Natural Language Processing and Machine Learning. However, there are many challenges and limitations in their application. In this survey we review recent advances on chatbots, where Artificial Intelligence and Natural Language processing are used. We highlight the main challenges and limitations of current work and make recommendations for future research investigation.
Despite the many advantages of computer-assisted data collection, it is unclear if, when, or under what conditions embodied conversational agents (i.e., ECA, virtual humans) can replace human ...interviewers to collect personal information in interviews, especially for topics that might be regarded as ‘sensitive’. This paper presents results from an exploratory study designed to investigate how topic sensitivity affects individuals' preference to disclose to a human or an ECA interviewer. A convenience sample of 203 undergraduate business students completed a scenario-based survey that asked them to rate the sensitivity of various interview topics and indicate their preference to disclose such sensitive information to human or ECA interviewers. Open-ended questions revealed factors behind preferences for interviewer choice. Findings show a preference for ECAs when topics are highly sensitive and more likely to evoke negative self-admissions. For topics rated low in sensitivity or more likely to evoke positive self-admissions, human interviewers are preferred. Specifically, participants stated that they would feel more comfortable discussing sensitive topics with an ECA interviewer because it could not judge them. This indicates that the evaluative capability of the interviewer plays a factor in the amount of sensitive information elicited from interviewees. Overall, results contribute to an understanding of when and why ECA interviewers can effectively replace human interviewers.
•Topic sensitivity affects individuals' preferences to disclose to a human or avatar.•Qualitative responses to open-ended questions about interviewer type were analyzed.•Study participants preferred avatar interviewers for more sensitive topics.•Study participants preferred human interviewers for less sensitive topics.•The ability to judge was the key factor differentiating interviewer preferences.
Personal interviews are widely used to collect open-ended data in an array of settings, many of which require solicitation of data that may be regarded as sensitive or threatening by interviewees. ...The high costs of employing human interviewers can be mitigated by utilizing computer-generated “human-like” interviewers, at least in scripted interview conditions. However, the effects of replacing humans with human-like interviewers on interviewees’ level of actual disclosure has not been well-researched. This experiment compared disclosure of sensitive information obtained from personal interviews utilizing various interviewer modes. One-hundred and fifty-eight students from a southwestern university were randomly assigned to answer a series of open-ended questions in one of three different personal interview modes: 1) audio-only computer-assisted self-interview (i.e., ACASI); 2) human-like embodied conversational agent (i.e., ECA); or 3) a human interviewer. Disclosure was measured using both self-report and objective scores derived from trained judges. Disclosure levels were significantly higher in the faceless ACASI condition than in the combined virtual and human interviewer conditions – both of which were embodied with a face. There was no significant difference between the ECA and human interviewer conditions. This suggests that the mere presence of a human face can inhibit disclosure.
•Presence of a human or human-like face decreases sensitive information disclosure.•No significant difference in disclosure to human or embodied conversational agent.•Choice to not disclose is often manifest by providing an irrelevant answer to the question.•No evidence that computerized interviewers perform worse than human interviewers.
This article reports on the second GENEA Challenge to benchmark data-driven automatic co-speech gesture generation. Participating teams used the same speech and motion dataset to build ...gesture-generation systems. Motion generated by all these systems was rendered to video using a standardised visualisation pipeline and evaluated in several large, crowdsourced user studies. Unlike when comparing different research articles, differences in results are here only due to differences between methods, enabling direct comparison between systems. The dataset was based on 18 hours of full-body motion capture, including fingers, of different persons engaging in a dyadic conversation. Ten teams participated in the challenge across two tiers: full-body and upper-body gesticulation. For each tier, we evaluated both the human-likeness of the gesture motion and its appropriateness for the specific speech signal. Our evaluations decouple human-likeness from gesture appropriateness, which has been a difficult problem in the field. The evaluation results show some synthetic gesture conditions being rated as significantly more human-like than 3D human motion capture. To the best of our knowledge, this has not been demonstrated before. On the other hand, all synthetic motion is found to be vastly less appropriate for the speech than the original motion-capture recordings. We also find that conventional objective metrics do not correlate well with subjective human-likeness ratings in this large evaluation. The one exception is the Fréchet gesture distance (FGD), which achieves a Kendall’s tau rank correlation of around -0.5. Based on the challenge results we formulate numerous recommendations for system building and evaluation.
Interventions to improve health and well-being abound. Whether they are designed for prevention, maintenance or improvement, a key challenge is the motivation of the user to change their current ...behaviours, such as persisting or taking new actions. To encourage someone to change their behaviour requires persuading them to change their goals and/or their beliefs about the behaviour or their ability to perform it. Our embodied conversational agent (ECA) uses explanations based on the goals and beliefs of the user to promote a sense of personalisation and engagement with the treatment plan which could form a bond as the dyad develop shared goals and tasks together. To keep our message minimal and understand whether belief-based or goal-based explanations are more efficacious in changing behaviour intention, we collected data in the context of a scenario where the ECA seeks to change four behaviours recommended to help students manage their study stress. Our findings suggest that when the behaviour requires a change in desire, we need goal-based explanation, when adoption of the behaviour requires addressing a barrier we need belief-based explanation and warrant future investigation. Further, the stratified analysis suggested that more tailoring to the student's context could provide more motivation to change.
In this paper we highlight the different challenges in modeling communicative gestures for Embodied Conversational Agents (ECAs). We describe models whose aim is to capture and understand the ...specific characteristics of communicative gestures in order to envision how an automatic communicative gesture production mechanism could be built. The work is inspired by research on how human gesture characteristics (e.g., shape of the hand, movement, orientation and timing with respect to the speech) convey meaning. We present approaches to computing where to place a gesture, which shape the gesture takes and how gesture shapes evolve through time. We focus on a particular model based on theoretical frameworks on metaphors and embodied cognition that argue that people can represent, reason about and convey abstract concepts using physical representations and processes, which can be conveyed through physical gestures.