Loading...
Please wait, while we are loading the content...
Similar Documents
What Can You Do ? Studying Social-Agent Orientation and Agent Proactive Interactions with a n Agent for Employees
| Content Provider | Semantic Scholar |
|---|---|
| Author | Liao, Q. Vera Davis, Matthew Geyer, Werner Muller, Michael Shami, N. Sadat |
| Copyright Year | 2016 |
| Abstract | Personal agent software is now in daily use in personal devices and in some organizational settings. While many advocate an agent sociality design paradigm that incorporates human-like features and social dialogues, it is unclear whether this is a good match for professionals who seek productivity instead of leisurely use. We conducted a 17-day field study of a prototype of a personal AI agent that helps employees find work-related information. Using log data, surveys, and interviews, we found individual differences in the preference for humanized social interactions (social-agent orientation), which led to different user needs and requirements for agent design. We also explored the effect of agent proactive interactions and found that they carried the risk of interruption, especially for users who were generally averse to interruptions at work. Further, we found that user differences in socialagent orientation and aversion to agent proactive interactions can be inferred from behavioral signals. Our results inform research into social agent design, proactive agent interaction, and personalization of AI agents. Author Keywords Agent; personalization; social-agent orientation; agent proactive interaction; enterprise personal agent. ACM Classification Keywords H.5.m. Information interfaces and presentation (e.g., HCI): Miscellaneous. INTRODUCTION Half a century after the introduction of first-generation chatbots like ELIZA [47], conversational agent interfaces have become increasingly common, as demonstrated by popular applications such as Apple Siri, Google Now and Microsoft Cortona. Many of these agents act as a new interface paradigm for information-finding, which incorporates or aims to replace the traditional interfaces such as search engines and recommender systems [12]. In addition to handling natural conversational interactions, scholars argue that the advantage of the agent inter-face lies in its social capabilities [1, 6, 7, 16, 21]. The social aspect of the agent interface not only provides potentially more engaging user experiences, but also engenders new technologies and designs that leverage a human metaphor and social context. For example, many systems utilize a “personal assistant” metaphor that makes users more receptive to functions such as reminders and task delegation [20, 31, 50]. Some embrace the opportunities for personalization by continuously learning about the user through social and relational conversations [16]. However, questions remain as to whether agent sociality, the incorporation of humanized social features, is favored for an information-finding application, the evaluation of which is often based on task performance such as accuracy and relevance of the information retrieved. The necessity of sociality may be especially questionable for users with high productivity versus leisurely needs, such as users in organizational settings. Moreover, we argue that there could be individual differences in preference for the level of agent sociality, potentially shaped by individual experiences with both the increasingly popular commercial agent applications and conventional information-finding applications such as search engines. Another potential of the agent interface, as some argue [5, 24, 52], is to initiate proactive interactions in a manner similar to how a person initiates conversations. Proactive interactions may serve a variety of purposes such as recommending, reminding [52], facilitating learning [24], or persuading [5]. However, proactive interactions may risk interrupting users. Especially for employees with busy schedules and thus scarce attention resources, interruption by agent proactive interactions may result in low responsiveness, and even worse, undermine the overall experience with the agent system. To explore these topics, we conducted a 17-day field study with a prototype of an enterprise personal agent that aims to help employees find work-related information in a largesize international company. We contribute to the humanagent interaction literature with a rare opportunity to study the social interactions and proactive interactions of agents Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions@acm.org. DIS 2016, June 04 08, 2016, Brisbane, QLD, Australia Copyright is held by the owner/author(s). Publication rights licensed to ACM. ACM 978-1-4503-4031-1/16/06 $15.00 DOI: http://dx.doi.org/10.1145/2901790.2901842 in a real user context for an extended period. The majority of user studies for agent systems rely on lab experiments (e.g. [17, 24, 48, 49]). Such a controlled environment and limited timeframe may not allow full exploration of the social aspects of human-agent interactions. Meanwhile, a few field studies documented that users exhibit anthropomorphic behaviors such as asking “how are you?” as well as flaming behaviors such as typing random letters [25, 46], which are not often observed in lab studies. Therefore, we believed a field study would be more suitable for studying agent sociality. Also, we studied an agent specialized in the enterprise context. While there is a growing interest in developing personal agents for the work environments [1, 18, 20, 52], there have been few field studies for these systems to the best of our knowledge. In this paper, we focused on studying individual differences in social-agent orientation, defined as the preference for humanized social interactions with an agent interface, such as having natural conversations and social dialogues. We explored how the individual differences led to differences in user requirements for agent system design and user behaviors that can be used to infer such orientation. By testing a variety of agent proactive interaction designs, including initiating social dialogue and crowdsourcing user questions, we also examined how users reacted to agent proactive interactions. Our study highlights that agent sociality designs are crucial for some users’ experience, but may not fit others, at least for a performance-driven, nonleisure context. We also underline the importance of considering user aversion to unsolicited proactive interactions in a work environment, especially for users with busy schedules and frequent social contacts. In the following section, we introduce the background of the study before discussing hypotheses and research questions. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://qveraliao.com/dis2016.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
