Resilient Chatbots

Ashktorab, Zahra; Jain, Mohit; Liao, Q. Vera; Weisz, Justin D.

doi:10.1145/3290605.3300484

Cited by 142 publications

(66 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As conversation is prone to breakdown, conversational repair is a critical capability of any conversationalist -human or machine [22]. [39] described conversational repair as a key conversational UX (user experience) pattern, and different approaches to conversational repair have been suggested [3]. Arguably, the most prevalent approach to conversational repair still is the fallback intent where a chatbot expresses failure to understand and invites the users to ask again.…”

Section: Chatbot Conversational Performancementioning

confidence: 99%

“…Conversational performance was operationalized in terms of the presence (or absence) of breakdown and repair (Section 2.3) for one of three tasks. Breakdown and repair followed the 'repeat' pattern of [3] where breakdown involved the chatbot failing to understand the user request and asking the user to reformulate, and repair entailed the chatbot's understanding of the users' reformulated request to provide a relevant response. Each variant was evaluated by different groups of participants (Section 3.3).…”

Section: Chatbot Designmentioning

confidence: 99%

See 1 more Smart Citation

Effects of Humanlikeness and Conversational Breakdown on Trust in Chatbots for Customer Service

Law

Følstad

As³

2022

Nordic Human-Computer Interaction Conference

View full text Add to dashboard Cite

Trust in chatbots can be shaped by various factors such as humanlikeness in terms of visual appearance and conversational content, and conversational performance in terms of the chatbot's ability to avoid conversational breakdown. The literature is inconclusive concerning the effect of humanlikeness and conversational performance on trust, especially their interaction effect. To examine the relations among these variables, we conducted a 2x3 (humanlikeness x conversational performance) factorial experiment with 251 participants, who were asked to perform three tasks with a chatbot for an online bank under one of the six conditions. Participants completed a questionnaire measuring trust and commented on trust factors. Results of between-group analysis showed that for the task with seeded breakdowns there were significant differences in trust across the six groups with the lowest ratings for the two groups experiencing breakdowns without repairs and that humanlikeness did not impact the extent to which the trust level changed. Results of within-group analysis showed significant differences in trust across the tasks but non-significant inter-task correlations on trust for the two groups. These observations challenge the effect of humanlikeness on trust while supporting the notion of trust resilience as the participants did not spill the impaired trust over the subsequent task. Thematic analysis showed that inter-group contrasts could be found for the theme 'underlying functionality' and 'affective responses.' Implications for research, practice and future work were drawn. CCS CONCEPTS• Human-centered computing → Empirical studies in HCI.

show abstract

Section: Chatbot Conversational Performancementioning

confidence: 99%

Section: Chatbot Designmentioning

confidence: 99%

Effects of Humanlikeness and Conversational Breakdown on Trust in Chatbots for Customer Service

Law

Følstad

As³

2022

Nordic Human-Computer Interaction Conference

View full text Add to dashboard Cite

show abstract

“…As has been stated before, the NLU module for the voice-based agent is managed directly by Amazon technology and therefore little work can be done on that by a third party [75]. On the other hand, the interaction model-i.e.…”

Section: Aida Cookbot's Interaction Modelmentioning

confidence: 99%

Diabetes and conversational agents: the AIDA project case study

Alloatti

Bosca²,

et al. 2021

Discov Artif Intell

View full text Add to dashboard Cite

One of the key aspects in the process of caring for people with diabetes is Therapeutic Education (TE). TE is a teaching process for training patients so that they can self-manage their care plan. Alongside traditional methods of providing educational content, there are now alternative forms of delivery thanks to the implementation of advanced Information Technologies systems such as conversational agents (CAs). In this context, we present the AIDA project: an ensemble of two different CAs intended to provide a TE tool for people with diabetes. The Artificial Intelligence Diabetes Assistant (AIDA) consists of a text-based chatbot and a speech-based dialog system. Their content has been created and validated by a scientific board. AIDA Chatbot—the text-based agent—provides a broad spectrum of information about diabetes, while AIDA Cookbot—the voice-based agent—presents recipes compliant with a diabetic patient’s diet. We provide a thorough description of the development process for both agents, the technology employed and their usage by the general public. AIDA Chatbot and AIDA Cookbot are freely available and they represent the first example of conversational agents in Italian to support diabetes patients, clinicians and caregivers.

show abstract

“…Liao et al [43] investigated how agent sociability influences user interactions with conversational agents. Ashktorab et al [4] studied preferences for different strategies to handle conversational breakdowns. Hearst et al [25,26] investigated the visual designs of answers provided by a natural language interface and how users perceive these designs.…”

Section: Question-answering Systems and User Perceptionmentioning

confidence: 99%

Causal Perception in Question-Answering Systems

Law

Endert

et al. 2021

Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Root cause analysis is a common data analysis task. While questionanswering systems enable people to easily articulate a why question (e.g., why students in Massachusetts have high ACT Math scores on average) and obtain an answer, these systems often produce questionable causal claims. To investigate how such claims might mislead users, we conducted two crowdsourced experiments to study the impact of showing different information on user perceptions of a question-answering system. We found that in a system that occasionally provided unreasonable responses, showing a scatterplot increased the plausibility of unreasonable causal claims. Also, simply warning participants that correlation is not causation seemed to lead participants to accept reasonable causal claims more cautiously. We observed a strong tendency among participants to associate correlation with causation. Yet, the warning appeared to reduce the tendency. Grounded in the findings, we propose ways to reduce the illusion of causality when using question-answering systems.

show abstract

Resilient Chatbots

Cited by 142 publications

References 31 publications

Effects of Humanlikeness and Conversational Breakdown on Trust in Chatbots for Customer Service

Effects of Humanlikeness and Conversational Breakdown on Trust in Chatbots for Customer Service

Diabetes and conversational agents: the AIDA project case study

Causal Perception in Question-Answering Systems

Contact Info

Product

Resources

About