Table 1 - uploaded by Nils Oliver Klowait
Content may be subject to copyright.
Source publication
The following study conceptualizes and evaluates a phone-based, natural-language-employing Automated Computer-Telephone Interviewing system. It will be argued that the conversational agent, by virtue of its technical limitations, is situated squarely within the interactional "uncanny valley," precisely because it exhibits a rudimentary interactivit...
Similar publications
The paper consists of two parts, i.e. theoretical and practical. The theoretical deliberations are devoted to the explanation of supporting the process of product co-creation. The practical part of this paper reports the findings of a study concerning the interest of young people in participation in co-creation in Poland and China. The aim of the a...
Citations
... First, there are smart assistants such as Alexa, Siri and Google Assistant (Alač et al. 2020;Fischer et al. 2019;Porcheron et al. 2017Porcheron et al. , 2018. Second, there are telephone systems such as Lenny that simulate real callers (Sahin et al. 2017;Relieu et al. 2020), and telephone systems that act as operators of some sort (Aranguren 2014;Avgustis et al. 2021;Wallis 2008;Wooffitt 1994) or do automated interviews (Klowait 2017). 12 "Wizard of Oz" is a setup where a human 'wizard' is, unbeknownst to the user, in control of the system's actions. ...
Despite its elusiveness as a concept, ‘artificial intelligence’ (AI) is becoming part of everyday life, and a range of empirical and methodological approaches to social studies of AI now span many disciplines. This article reviews the scope of ethnomethodological and conversation analytic (EM/CA) approaches that treat AI as a phenomenon emerging in and through the situated organization of social interaction. Although this approach has been very influential in the field of computational technology since the 1980s, AI has only recently emerged as such a pervasive part of daily life to warrant a sustained empirical focus in EM/CA. Reviewing over 50 peer-reviewed publications, we find that the studies focus on various social and group activities such as task-oriented situations, semi-experimental setups, play, and everyday interactions. They also involve a range of participant categories including children, older participants, and people with disabilities. Most of the reviewed studies apply CA’s conceptual apparatus, its approach to data analysis, and core topics such as turn-taking and repair. We find that across this corpus, studies center on three key themes: openings and closing the interaction, miscommunication, and non-verbal aspects of interaction. In the discussion, we reflect on EM studies that differ from those in our corpus by focusing on praxeological respecifications of AI-related phenomena. Concurrently, we offer a critical reflection on the work of literature reviewing, and explore the tortuous relationship between EM and CA in the area of research on AI.
... Conversely, if conversational flexibility is desired -this is especially true for ECAs that try to maximize realism and generalize its range of applications -the more flexible synthetic voice is preferable, especially since conversational interactivity can moderate the de-anthropomorphizing effect of robotic voice. A negative example of the latter case is described in Klowait [2017], where an automated computer-telephone interviewing system was created with realistic pre-recorded human voices, yet integrated with a contextually inflexible conversational system. The resulting negative user experiences -a considerable number of which never reached understanding that they were talking to a limited AI rather than a rude/incapable human -could have been avoided by opting for an obviously synthetic voice. ...
... In particular, studies of interaction with conversational agents reveal problems with the turn-taking system (transfer of speakership from an ECA to a person and vice versa): people either start talking too early or 'skip' the allotted space for an answer, as a result of which their utterance is ignored, forcing participants repeat the sequence of interaction over and over [Arend, Sunnen, 2017;Pelikan, Broth, 2016;Pitsch, Gehle, Dankert, Wrede, 2017]. Furthermore, conversational agents may overlook the constant [Klowait, 2017;Trott, Rossano, 2017]. ...
The field of human-computer interaction (HCI) investigates the intersection between the design of devices and users. From an early focus on interaction modeling based on psychological experiments, the field has since experienced a shift towards the study of how actual humans interact with autonomous devices. The field became conductive to ethnographic, observational and videographic studies of human-device interaction. Conversation-analytic HCI became possible. That said, this new wave of researchers was never truly able to dethrone the psychological common sense of the field. With recent developments in both the technical-sensorial capabilities and outward actuational range of embodied virtual agents, the field of HCI has once again returned to the question of the sequential unfolding of the interaction between users and intelligent agents, and the multimodal interactional repertoire that is deployed throughout. This review will highlight the situational orientation of high-impact research in the field, and relate it to the cotemporaneous development of ethnomethodological and conversation analytic frameworks.
Acknowledgments. The article was prepared in the framework of a research grant funded by the Ministry of Science and Higher Education of the Russian Federation (grant ID: 075-15-2020-908). The article was prepared in cooperation with the Sber (ex. – Sberbank’s) Gamification Lab.
... In terms of HCI research, the test environment construction is maximally complex, as it has to replicate mediated human interaction (Brahnam and de Angeli 2012;Sundar et al. 2014;Klowait 2017). If the first approach had as its ultimate end the android, this approach strives towards a complete virtual intelligence. ...
Nass’ and Reeves’ media equation paradigm within human–computer interaction (HCI) challenges long-held assumptions about how users approach computers. Given a rudimentary set of cues present in the system’s design, users are said to unconsciously treat computers as genuine interactants—extending rules of politeness, biases and human interactive conventions to machines. Since the results have wide-ranging implications for HCI research methods, interface design and user experiences, researchers are hard-pressed to experimentally verify the paradigm. This paper focuses on the methodology of attributing the necessary social cues to the agent, a core aspect of the experimental design of studies dealing with the media equation. A typology of experimental anthropomorphisms is developed, allowing an assessment of how the differing axiomatic assumptions affect the relevance of the results for an evaluation of the paradigm. The paper concludes with a series of arguments in favour of one particular anthropomorphism type for researching the media equation.
Разговорные компьютеризированные агенты совершенствуются, развиваются и получают все более широкое распространение в различных сферах жизни. Теперь искусственный интеллект может вести беседу с человеком, используя сложные разговорные приемы и человекоподобные речевые паттерны. Например, Google Duplex, современный голосовой интерфейс, способен автономно выполнять сервисные телефонные звонки таким образом, что собеседнику не удается опознать в нем робота.
С увеличением числа интерактивных коммуникаций человека и компьютера становится очевидной необходимость их изучения: каким образом они происходят, как разрешаются недоразумения и достигаются цели разговора. Статья посвящена эмпирическому изучению взаимодействия между продвинутым разговорным агентом и человеком: в рамках этнометодологии и конверсационного анализа рассмотрены последовательно разворачивающиеся, естественно возникающие речевые практики собеседников, а также порождаемые ими нормативные структуры.
Материалом для анализа послужила аудиозапись телефонной беседы между Google Duplex и человеком-оператором. Исследовано, в частности, как достигается разговорное закрытие — микроструктура, требующая кооперации собеседников в организации чередования реплик. Показано, что завершение беседы происходит за счет постепенного достижения эпистемического равновесия между говорящими, несмотря на интеракционные проблемы, возникающие из-за неуместного просодического контура робота. Подчеркивается связь между институциональным контекстом взаимодействия и роботизированной природой не-человеческого собеседника.
Полученные результаты закладывают основу для дальнейших исследований специфического порядка во взаимодействии человека и робота и способствуют более глубокому пониманию того, как завершается разговор в пограничных случаях.
Благодарности. Статья подготовлена в сотрудничестве с Лабораторией геймификации Сбербанка.
Статья подготовлена в рамках гранта, предоставленного Министерством науки и высшего образования Российской Федерации (№ соглашения о предоставлении гранта: 075-15-2020-908).
The article analyzes how an emerging form of automation may drastically transform contemporary employment dynamics. Recent breakthroughs in the field of artificial intelligence (AI) make it possible to automate both manual and mental non-standard tasks. The first part of the article traces the development of AI. Whereas classical algorithms required the creation of a hermetic environment for AI to thrive, modern neural network-based AI is capable of surviving in the chaotic realm occupied by humans. Based on an analysis of changes in the nature of AI, the authors distinguish between substitutive and supplemental automation. The former refers to a complete replacement of humans by machines, while the latter indicates a selective substitution of humans in specific professional functions. In order to conceptualize professions as a nexus of automatable components, the authors employ Goffman’s dramaturgical framework. Goffman studied the social visibility of professional activity. Goffman held that any profession can be divided into invisible routines that are fundamental to it and a dramatization that makes the profession socially visible. The article demonstrates that the current utopian and anti-utopian views of automation both reduce work to its visible components and neglect the logic of supplemental automation. The authors argue that the targets of modern automation are not the socially visible components but the invisible routines. In the final section, the authors develop a model that takes these invisible professional routines into account and analyze what effect this new type of automation may have on different types of professions with differing degrees of social visibility.
This article examines the methodological difficulties surrounding the topic of agency-ascription in human-computer interaction (HCI). The author evaluates a certain shift which occurred within this field, from analyzing computer-human interactions as model-like abstractions, which could be expressed in "program-like" scripts, to analyzing human-computer interactions as situated and locally unfolding events. This shift towards the situational, which has occurred through conversation-analytic interventions by authors such as L. Suchman and P. Luff, et al., raised the possibility of a conversation-analytic analysis of human-computer interactions. However, the use of the conceptual repertoire of conversation analysis poses a dilemma: the "talk is action" paradigm of CA would demand a certain conception of interaction, where all interactional participants would have some degree of agency, including the lifeless computer with its predetermined behavioral scripts. The paper proceeds to investigate one of the supposed solutions to the action vs. system dilemma: instead of stipulating that the computer possesses "true" agency, CA-minded researchers may instead opt for an "agency ascription" on behalf of the human co-interactant. This anthropomorphism would supposedly permit an interactionist analysis without paying the price of accepting a non-human ontology. The paper looks into two ideal types of anthropomorphism which are present in interaction-focused HCI research: ontological and pragmatic anthropomorphism, analyzing their distinct features and how they relate to the possibility of analyzing HCI situationally. C. Nass' "Media Equation" paradigm is considered as an example of ontological anthropomorphism, while S. Barley's analysis of the introduction of CAT-scanners in two hospitals is taken as the foundation towards a pragmatic conception of agency-ascription, which is crucially dependent on the concepts of "seriousness" and "literality" as they occur in situ. The paper concludes by cautioning against the use of anthropomorphism as a justification for interactionist HCI. It argues that the researcher cannot effectively probe for "serious" agency-ascription, as the degree of seriousness depends on the situation. © 2018 Institute of Sociology of the Russian Academy of Sciences. All rights reserved.
