Generative AI has caused a paradigm shift in the area of Artificial Intelligence (AI) and as such has inspired much new research, especially on Large Language Models (LLMs). LLMs are transforming how people interact with computers in service-oriented fields in both the consumer (for example: retail, travel, education, healthcare) and enterprise (customer care, field service, sales, marketing, etc.) spaces. One barrier to widespread adoption is the current unpredictability of LLM behavior: users must trust that LLM-based services and systems are accurate, fair, and unbiased. Model responses that exhibit biases related to race, social status, and other sensitive topics can have serious consequences, ranging from lack of trust in the model to adverse social implications for consumers, all the way to damage to the reputations of the corporations that provide them. This study explores how to uncover biases related to social stigmas in LLM output, by using an adversarial prompt-based approach. Discovering model vulnerabilities of this type is a nontrivial task due to the large search space, making it resource-intensive. We present an evaluation framework for probing and analyzing the behaviors of multiple LLMs systematically. We use a curated set of adversarial prompts with a focus on uncovering biased responses to prompts associated with social attributes.

Although methods for repairing prior turns in natural conversation are critical for enabling mutual understanding, or successful communication, these methods are seldom built into conversational user interfaces systematically. Chatbots and voice assistants tend to ask users to paraphrase what they said if it was not understood, but users cannot do the same if they encounter trouble in understanding what the agent said. Understanding is a one-way street in most (intent-based) conversation-like interfaces. An exception to this is Moore and Arar (2019), who demonstrate nine types of user-initiated repair on agent responses that are common in natural conversation and who have shown that users will employ these repair features correctly in text-based interfaces if taught. In this small-scale study, we test these user-initiated repairs (in second position) in a voice-based interface. With understanding-oriented repairs, we found that participants employed them much the same way in text and voice. In addition, we examine some hearing- and speaking-oriented repairs that emerged from the use of our novel multi-modal interface. We found that participants used them to manage troubles specific to the voice modality. Analysis of user logs and transcripts suggests that user-initiated repair features are valuable components of conversational interfaces.

User interfaces that take human conversation as their interaction metaphor work fundamentally differently than those that employ spatial metaphors, such as a desktop or a page. While the fundamenta...

Conversational agents (CAs) are systems that interact with humans through natural language user interfaces. They include systems with a range of conversational capabilities and modalities. For example, there are text- or voice-only question-answering interactions such as Apple Siri, Google Assistant, and Amazon Alexa, and there are also multimodal conversational AI agents that can engage users in long-term dialogues. Advances in speech recognition, natural language processing, and computer vision have resulted in a greater acceptance and use of CAs. CAs have already started to play important roles in various healthcare settings, including assisting clinicians during consultations, assisting consumers in changing health behaviours, and helping patients such as the elderly in their living environments. \n \nThere have been several systematic reviews on the use of CAs in health and wellbeing recently. Although the field still appears to be nascent, the emerging evidence has shown user acceptance of CAs in the healthcare domain as well as the early promises in boosting healthcare outcomes in both physical and mental health. Despite the increasing adoption and the benefits of using CAs to support health and wellbeing, the review studies also revealed (i) patient safety was rarely examined, (ii) health outcomes were inadequately measured, and (iii) no standardised evaluation methods were employed. There were also limitations in reporting the technical implementation details of CAs used, making the replicability of prior studies problematic. \n \nIn addition to addressing some of the current challenges and limitations, this special issue features cutting-edge research on designing, developing, and evaluating CAs for health and wellbeing that aim to improve health outcomes and services, and satisfy unique application needs (e.g., safety, trust, and user experience). The seven articles included in this special issue cover many application areas ranging from mental health and social support to information seeking to coaching. Amongst the accepted articles, mental health and social support themes represented the primary research foci. The articles also covered different population groups including older adults, young adults, and homeless people. Based on their foci, we have grouped the articles in this special issue by three areas: mental health, older adult wellbeing, and social support and coaching.

Domain-specific knowledge bases (KB), carefully curated from various data sources, provide an invaluable reference for professionals. Conversation systems make these KBs easily accessible to professionals and are gaining popularity due to recent advances in natural language understanding and AI. Despite the increasing use of various conversation systems in open-domain applications, the requirements of a domain-specific conversation system are quite different and challenging. In this paper, we propose an ontology-based conversation system for domain-specific KBs. In particular, we exploit the domain knowledge inherent in the domain ontology to identify user intents, and the corresponding entities to bootstrap the conversation space. We incorporate the feedback from domain experts to further refine these patterns, and use them to generate training samples for the conversation model, lifting the heavy burden from the conversation designers. We have incorporated our innovations into a conversation agent focused on healthcare as a feature of the IBM Micromedex product.

Business intelligence (BI) applications play an important role in the enterprise to make critical business decisions. Conversational interfaces enable non-technical enterprise users to explore their data, democratizing access to data significantly. In this paper, we describe an ontology-based framework for creating a conversation system for BI applications termed as Conversational BI. We create an ontology from a business model underlying the BI application, and use this ontology to automatically generate various artifacts of the conversation system. These include the intents, entities, as well as the training samples for each intent. Our approach builds upon our earlier work, and exploits common BI access patterns to generate intents, their training examples and adapt the dialog structure to support typical BI operations. We have implemented our techniques in Health Insights (HI) , an IBM Watson Healthcare offering, providing analysis over insurance data on claims. Our user study demonstrates that our system is quite intuitive for gaining business insights from data. We also show that our approach not only captures the analysis available in the fixed application dashboards, but also enables new queries and explorations.

Internet search engines display understanding or misunderstanding of user intent in and through the particular batches of results they retrieve and their perceived relevance. Yet understanding is not simply an automatic outcome but a joint interactional achievement between human and machine. If potential troubles with search queries emerge, either the user or the search engine may initiate repair on the query in ways that resemble repair in human conversation as described in conversation analysis. Users can repair their own queries in first or third position, while search engines can initiate repair from second position. However search-engine interactions currently contain no fourth-position repair. Finally search engines may also complete queries collaboratively with users in ways that are similar to but distinct from repair. In this study we examine interactions between users and search engines using a novel approach we call "computer interaction analysis," which utilizes eye-tracking screen video and a novel notation scheme for transcribing it.

Media research has shown that people enjoy watching television as a part of socializing in groups. However, many constraints in daily life limit the opportunities for doing so. The Social TV project builds on the increasing integration of television and computer technology to support sociable, computer-mediated group viewing experiences. In this article, we describe the initial results from a series of studies illustrating how people interact in front of a television set. Based on these results, we propose guidelines as well as specific features to inform the design of future “social television” prototypes.

Massively multiplayer online games (MMOGs) can be fascinating laboratories to observe group dynamics online. In particular, players must form persistent associations or "guilds" to coordinate their actions and accomplish the games' toughest objectives. Managing a guild, however, is notoriously difficult and many do not survive very long. In this paper, we examine some of the factors that could explain the success or failure of a game guild based on more than a year of data collected from five World of Warcraft servers. Our focus is on structural properties of these groups, as represented by their social networks and other variables. We use this data to discuss what games can teach us about group dynamics online and, in particular, what tools and techniques could be used to better support gaming communities.

Massively multiplayer online games (MMOGs) currently represent the most widely used type of social 3D virtual worlds with millions of users worldwide. Although MMOGs take face-to-face conversation as their metaphor for user-to-user interaction, avatars currently give off much less information about what users are doing than real human bodies. Consequently, users routinely encounter slippages in coordination when engaging in joint courses of action. In this study, we analyze screen-capture video of user-to-user interaction in the game, City of Heroes, under two conditions: one with the game's standard awareness cues and the other with enhanced cues. We use conversation analysis to demonstrate interactional slippages caused by the absence of awareness cues, user practices that circumvent such limitations and ways in which enhanced cues can enable tighter coordination.

Massively Multiplayer Online Games (MMOGs) routinely attract millions of players but little empirical data is available to assess their players' social experiences. In this paper, we use longitudinal data collected directly from the game to examine play and grouping patterns in one of the largest MMOGs: World of Warcraft. Our observations show that the prevalence and extent of social activities in MMOGs might have been previously over-estimated, and that gaming communities face important challenges affecting their cohesion and eventual longevity. We discuss the implications of our findings for the design of future games and other online social spaces.

Playing computer games has become a social experience. Hundreds of thousands of players interact in massively multiplayer online games (MMORPGs), a recent and successful genre descending from the pioneering multi-user dungeons (MUDs). These new games are purposefully designed to encourage interactions among players, but little is known about the nature and structure of these interactions. In this paper, we analyze player-to-player interactions in two locations in the game Star Wars Galaxies. We outline different patterns of interactivity, and discuss how they are affected by the structure of the game. We conclude with a series of recommendations for the design and support of social activities within multiplayer games.