Future directions for chatbot research: an interdisciplinary research agenda

In the paper

“In response, chatbots have emerged as a substantial research area in recent years.”

• Also stated in: § Introduction · plain

  “There has lately been a marked increase of interest in chatbot research within academia and industry, specifically from 2016 and onwards [86].”

What the sources say

• Not checked yetcites [86] Piccolo LS, Mensio M, Alani H (2018) Chasing the chatbots. In Internet Science. INSCI 2018 international workshops. Springer, Cham, pp 157–169 source

In the paper

“The rapidly growing body of chatbot research has a marked interdisciplinary character—spanning fields such as informatics, management and marketing, media and communication science, linguistics and philosophy, psychology and sociology, engineering, design, and human-computer interaction.”

• Also stated in: § Research communities · boosted

  “Chatbot research is currently evolving within and across a range of disciplines and has a strong interdisciplinary character.”

In the paper

“This has helped identify factors contributing to positive or negative user experience [118] and has addressed specific aspects such as trust [119], perceived social support [71], human likeness [4], and how these aspects are impacted by chatbot design [42].”

What the sources say

• Not checked yetcites [118] Zarouali B, Van den Broeck E, Walrave M, Poels K (2018) Predicting consumer responses to a chatbot on Facebook. Cyberpsychol Behav Soc Netw 21(8):491–497 source
• Not checked yetcites [119] Zierau N, Engel C, Söllner M, Leimeister JM (2020) Trust in smart personal assistants: A systematic literature review and development of a research agenda. In: Proceedings of the 15th international conference on Wirtschaftsinformatik—WI2020. Association for Information Systems,pp 99–11 source
• Not checked yetcites [71] Lee S, Choi J (2017) Enhancing user experience with conversational agent for movie recommendation: effects of self-disclosure and reciprocity. Int J Hum Comput Stud 103:95–105 source

In the paper

“Users still experience issues in chatbot interaction, both in terms of pragmatic experiences—where chatbots fail to understand or to help users achieve their intended goals [75]—and in terms of hedonic experiences—where chatbots fail to engage users over time [117].”

What the sources say

• Not checked yetcites [75] Luger E, Sellen A (2016) “Like having a really bad PA”—the gulf between user expectation and experience of conversational agents. In: Proceedings of CHI 2016. ACM, New York, pp 5286–5297 source
• Not checked yetcites [117] Zamora J (2017) I’m sorry, Dave, I’m afraid I can’t do that: Chatbot perception and expectations. In: Proceedings of the 5th international conference on human agent Interaction. ACM,New York,pp 253–260 source

In the paper

“Chatbot user experience has been a key theme in recent research efforts, for voice-based virtual agents [75] and text-based applications [4].”

What the sources say

• Not checked yetcites [75] Luger E, Sellen A (2016) “Like having a really bad PA”—the gulf between user expectation and experience of conversational agents. In: Proceedings of CHI 2016. ACM, New York, pp 5286–5297 source
• Not checked yetcites [4] Araujo T (2018) Living up to the chatbot hype: the influence of anthropomorphic design cues and communicative agency framing on conversational agent and company perceptions. Comput Hum Behav 85:183–189 source

In the paper

“Through the CONVERSATIONS workshop series, six overarching topics for future chatbot research have been identified.”

In the paper

“Evaluation in chatbot research is conducted by instruments for users' self reports of user experience [63], through user observation and interviews [75] and analyses of chatbot interaction [66], and also by physiological measurements [22].”

What the sources say

• Not checked yetcites [63] Kocaballi AB, Laranjo L, Coiera E (2019) Understanding and measuring user experience in conversational interfaces. Interact Comput 31(2):192–207 source
• Not checked yetcites [75] Luger E, Sellen A (2016) “Like having a really bad PA”—the gulf between user expectation and experience of conversational agents. In: Proceedings of CHI 2016. ACM, New York, pp 5286–5297 source
• Not checked yetcites [66] Kvale K, Sell OA, Hodnebrog S, Følstad A (2019) Improving conversations: lessons learnt from manual analysis of chatbot dialogues. In: Chatbot research and design. Third international workshop CONVERSATIONS 2019. Springer, Cham, pp 187–200 source

In the paper

“While we have substantial knowledge on current chatbot users, important topics lack sufficient coverage. Two warrant particular mention: (a) broader chatbot uses and user groups and (b) implications of chatbot use, both detailed below.”

In the paper

“Chatbots are increasingly becoming important gateways to digital services and information—taken up within domains such as customer service, health, education, and work support.”

• Also stated in: § Introduction · plain

  “Chatbots are currently taken up in application areas as diverse as customer service [1], health [105], education [53], and office work [78].”

What the sources say

• Not checked yetcites [1] Adam M, Wessel M, Benlian A (2020) AI-based chatbots in customer service and their effects on user compliance. Electron Markets. https://doi.org/10.1007/s12525-020-00414-7 source
• Not checked yetcites [105] Vaidyam AN, Wisniewski H, Halamka JD, Kashavan MS, Torous JB (2019) Chatbots and conversational agents in mental health: a review of the psychiatric landscape. Can J Psychiatry 64(7):456–464. https://doi.org/10.1177/0706743719828977 source
• Not checked yetcites [53] Hobert S, Berens F (2019) Small talk conversations and the long-term use of chatbots in educational settings–experiences from a field study. In: Chatbot research and design. Third international workshop CONVERSATIONS 2019. Springer, Cham, pp 260–272 source

In the paper

“For example, in 2019, over 50 % of US and German consumers were estimated to have used chatbots at least once—with even higher numbers in the UK or France [88].”

• Also stated in: § Users and implications · plain

  “Chatbot use is becoming commonplace.”

What the sources say

• Not checked yetcites [88] Press G (2019) AI Stats News: 62 % of US Consumers like using chatbots to interact with businesses. Forbes. Retrieved from https://www.forbes.com/sites/gilpress/2019/10/25/ai-stats-news-us-consumers-interest-in-using-chatbots-to-interact-with-businesses-rise-to-62/

In the paper

“we lack the needed technological underpinnings to support some key aspects of chatbot applications.”

In the paper

“it is promising to see that general guidelines for human-AI interaction are emerging [3] and more of these are needed.”

What the sources say

• Not checked yetcites [3] Amershi S, Weld D, Vorvoreanu M, Fourney A, Nushi B, Collisson P, Teevan J (2019) Guidelines for human–AI interaction. In: Proceedings of the CHI 2019 (paper no. 3). ACM, New York source

In the paper

“While the current research on chatbots typically addresses dyadic interactions between one chatbot and one user”

In the paper

“along with high-quality open-source libraries [113] have lowered the barrier to entry and make it possible to build conversational models that exhibit high generalization and coherence [90].”

What the sources say

• Not checked yetcites [113] Wolf T, Debut L, Sanh V, Chaumond J, Delangue C, Moi A, Cistac P, Rault T, Louf R, Funtowicz M, Brew J (2019) HuggingFace’s transformers: state-of-the-art natural language processing. arXiv preprint, arxiv:1910.03771 source
• Not checked yetcites [90] Roller S, Dinan E, Goyal N, Ju D, Williamson M, Liu Y, Boureau YL (2020) Recipes for building an open-domain chatbot. arXiv preprint. arXiv:2004.13637 source

In the paper

“The advances in chatbot frameworks and platforms are key enablers of the current interest in chatbot applications.”

In the paper

“There is an essential challenge in studying and designing chatbots for collaboration due to the multifaceted character of such interaction, and the range of potential theoretical perspectives to apply.”

In the paper

“human-chatbot teams which handle challenging service queries better than chatbots alone and more efficiently than humans alone.”

What the sources say

• Not checked yetcites [45] Grudin J, Jacques R (2019) Chatbots, humbots, and the quest for artificial general intelligence. In: Proceedings of CHI 2019 (paper no. 209). ACM, New York source

In the paper

“further progress in the field of chatbots will depend on advances in natural language understanding, which will remain a key area of research interest.”

In the paper

“criticism around large models has been growing lately [9], especially ethical concerns regarding undesirable and often inscrutable societal biases percolating the models [9, 120], carbon footprint [9, 99], misuse and misinterpretation [9], privatization of AI research [99], and even research opportunity costs [49].”

What the sources say

• Not checked yetcites [9] Bender EM, Gebru T, McMillan-Major A, Mitchell M (2021) On the dangers of stochastic parrots: can language models be too big? In: Proceedings of the 2021 ACM conference on fairness, accountability, and transparency. ACM, New York, pp 610–623 source
• Not checked yetcites [120] Zou J, Schiebinger L (2018) AI can be sexist and racist—it’s time to make it fair. Nature 559:324–326 source
• Not checked yetcites [99] Strubell E, Ganesh A, McCallum A (2019) Energy and policy considerations for deep learning in NLP. arXiv preprint, arXiv:1906.02243 source

In the paper

“The deliberative process at the workshop series was key to identify and propose research directions in a true interdisciplinary fashion.”

• Also stated in: § Approach · plain

  “As a result, the proposed research directions reflect the interdisciplinary position of a group of collaborating researchers within this emerging field.”

In the paper

“the introduction of the Transformer [107] as a dominant and highly effective architecture for natural language processing”

What the sources say

• Not checked yetcites [107] Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, Kaiser Ł, Polosukhin I (2017) Attention is all you need. In: Conference on neural information processing systems—NIPS 2017, pp 5998–6008

In the paper

“the use of generative approaches has seen substantial progress, where end-to-end dialogue systems are applied to predict suitable responses to user input based on models built from large conversational datasets [2, 109].”

What the sources say

• Not checked yetcites [2] Adiwardana D, Luong MT, So DR, Hall J, Fiedel N, Thoppilan R, Le QV (2020) Towards a human-like open-domain chatbot. arXiv preprint. arXiv:2001.09977
• Not checked yetcites [109] Vinyals O, Le Q (2015) A neural conversational model. arXiv preprint, arXiv:1506.05869

In the paper

“Third, challenges remain in solutions for supporting chatbot development and standardised testing for example in terms of studies simulating production environments and approaches to improve chatbots more easily in production.”

In the paper

“Thus, opening up these powerful models to the public has the potential to accelerate chatbot development even further.”

In the paper

“Key advances include the application of supervised machine learning for classification and information retrieval—enabling, for example, intent prediction and identification of user sentiment [20], which are critical to support task-oriented conversations.”

What the sources say

• Not checked yetcites [20] Chen H, Liu X, Yin D, Tang J (2017) A survey on dialogue systems: recent advances and new frontiers. ACM SIGKDD Explor Newsl 19(2):25–35 source

In the paper

“voice-based conversational agents have been shown to more easily interpret particular English dialects potentially reducing their utility for users of specific areas [51]”

What the sources say

• Not checked yetcites [51] Harwell D (2018) The accent gap. The Washington Post. Retrieved from https://www.washingtonpost.com/graphics/2018/business/alexa-does-not-understand-your-accent/

In the paper

“Current efforts are already moving in that direction, providing development resources that promise anyone with enough motivation, regardless of their background, to deliver human-like interactive experiences.”

In the paper

“Chatbots will both raise critical ethical challenges and hold implications for democratization of technology”

In the paper

“we foresee that chatbots in collaborative relations involving more people and bots will become more prominent as chatbots mature further.”

In the paper

“According to these guidelines, it is of paramount importance for trustworthy AI to be aligned with (a) legal regulations and (b) ethical principles and values, and also (c) be robust from a technical perspective given its particular social context.”

What the sources say

• Not checked yetcites [30] European Commission (2019) Ethics guidelines for trustworthy AI. Retrieved from: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai

In the paper

“Chatbots have been suggested as accessible interactive systems for visually impaired in need of an easily navigable user interface [7], as conversational support for users with special needs [19], and to support youth to engage in societal issues [110].”

What the sources say

• Not checked yetcites [7] Baez M, Daniel F, Casati F (2019) Conversational web interaction: proposal of a dialog-based natural language interaction paradigm for the Web. In: Chatbot research and design. Third international workshop, CONVERSATIONS. Springer, Cham, pp 94–11 source
• Not checked yetcites [19] Catania F, Di Nardo N, Garzotto F, Occhiuto D (2019) Emoty: an emotionally sensitive conversational agent for people with neurodevelopmental disorders. In Proceedings of the 52nd Hawaii international conference on system sciences. Scholar Space, Honolulu, pp 2014–2023 source
• Not checked yetcites [110] Väänänen K, Hiltunen A, Varsaluoma J, Pietilä I (2019) CivicBots–Chatbots for supporting youth in societal participation. In: Chatbot research and design. Third international workshop CONVERSATIONS 2019. Springer, Cham, pp 143–157 source

In the paper

“based on the advice of a high-level expert group on AI, a European set of ethics guidelines for trustworthy AI has been presented [30].”

What the sources say

• Not checked yetcites [30] European Commission (2019) Ethics guidelines for trustworthy AI. Retrieved from: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai

In the paper

“This concern for ethics in AI is motivated by its disruptive character and potential for changes in the job market and misuse by malevolent actors, as well as issues pertaining to accountability and bias [58].”

What the sources say

• Not checked yetcites [58] McTear MF (2002) Spoken dialogue technology: enabling the conversational user interface. ACM Comput Surv (CSUR) 34(1):90–169 source

In the paper

“while available platforms and frameworks are promoted as low-threshold means of chatbot design and development, there is a lack of knowledge regarding how these are actually employed to democratize chatbot development and design.”

In the paper

“tasks assigned to chatbots in these collaborative interactions can vary in complexity and involvement.”

In the paper

“Such tasks can be as simple as providing individual notifications, or as complicated as communicating processed and analysed data to different stakeholders.”

In the paper

“a proposed European set of regulations for AI, the AI Act, will help strengthen aspects of ethical concern in AI systems, including legal requirements for human oversight, accuracy, robustness, and security.”

In the paper

“as noted in a review of the chatbot literature, there has been an initial lack of ethical discussion in chatbot research [102]—though noteworthy exceptions to this exist, such as the exploration of ethical and social considerations for conversational AI by Ruane et al. [91].”

What the sources say

• Not checked yetcites [102] Syvänen S, Valentini C (2020) Conversational agents in online organization–stakeholder interactions: a state-of-the-art analysis and implications for further research. J Commun Manag 339–362 source
• Not checked yetcites [91] Ruane E, Birhane A, Ventresque A (2019) Conversational AI: social and ethical considerations. In: Proceedings for the 27th AIAI Irish conference on artificial intelligence and cognitive science—AICS 2019, pp 104–115. CEUR Workshop Proceedings, Vol-2563. http://ceur-ws.org/Vol-2563/

In the paper

“Twitter bots utilizing COVID-19 content to spread political conspiracies [34]”

What the sources say

• Not checked yetcites [34] Ferrara E (2020) What types of COVID-19 conspiracies are populated by Twitter bots? First Monday. https://doi.org/10.5210/fm.v25i6.10633 source

In the paper

“due to technological limitations and design choices, they can be the means to perpetuate and even reinforce existing biases in society, exclude or discriminate against some user groups (e.g. [33, 51]) and over-represent or enshrine specific values”

What the sources say

• Not checked yetcites [33] Feine J, Gnewuch U, Morana S, Maedche A (2019b) Gender bias in chatbot design. In: Chatbot research and design. Third international workshop CONVERSATIONS 2019. Springer, Cham, pp 79–93 source
• Not checked yetcites [51] Harwell D (2018) The accent gap. The Washington Post. Retrieved from https://www.washingtonpost.com/graphics/2018/business/alexa-does-not-understand-your-accent/

In the paper

“By allowing simple natural language dialogues, chatbots are potentially a low-threshold means to access information and services and may as such serve to bridge digital divides and strengthen inclusion [14].”

What the sources say

• Not checked yetcites [14] Brandtzaeg PB, Følstad A (2017) Why people use chatbots. In: Proceeedings of the international conference on internet science—INSCI 2017. Springer, Cham, pp 377–392 source

In the paper

“AI has recently been the objective of substantial interest from policy-making and regulatory bodies, as well as in discussions and reflections on ethics, privacy management and trust [21].”

What the sources say

• Not checked yetcites [21] Chung H, Iorga M, Voas J, Lee S (2017) Alexa, can I trust you? Computer 50(9):100–104 source

In the paper

“the General Data Protection Regulation (GDPR) is now used to govern privacy in technology-based systems and services.”

In the paper

“consequently are a potential low threshold means to access information and services and promote inclusion”

In the paper

“The aim of using chatbots for strengthening democratization, reducing bias, and facilitating universal design has been included in the vision of chatbots for social good [40]”

What the sources say

• Not checked yetcites [40] Følstad A, Brandtzaeg PB, Feltwell T, Law ELC, Tscheligi M, Luger E (2018) Chatbots for social good. In: Extended abstracts of the 2018 CHI conference on human factors in computing systems (paper no. SIG06). ACM, New York

In the paper

“While this problem of human–machine collaboration is addressed in more generic terms, for example in actor-network theory [70], there is a lack of models to characterize conversational collaboration involving agents.”

What the sources say

• Not checked yetcites [70] Latour B (2005) Reassembling the Social - An introduction to actor-network-theory. Oxford University Press, Oxford, UK source

In the paper

“From these principles, the European Commission expert group has identified seven key requirements for ethical AI applications, including human agency and oversight, privacy and data governance, and diversity, non-discrimination, and fairness.”

What the sources say

• Not checked yetcites [30] European Commission (2019) Ethics guidelines for trustworthy AI. Retrieved from: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai

In the paper

“the conversational character of chatbots may conceivably introduce a range of specific ethical problems, for example the ethical implications of human-like and socially present chatbot interaction, issues of consent, the privacy implications of third-party interactions and the implications of emotional effect on children and vulnerable users.”

In the paper

“Chatbots will both raise critical ethical challenges and hold implications for democratization of technology”

In the paper

“there is a lack of more systematic or structured investigations of universal and inclusive design of chatbots.”

In the paper

“the general trend of deploying bots in large scale for political interference and influence [44]”

What the sources say

• Not checked yetcites [44] Gorwa R, Guilbeault D (2020) Unpacking the social media bot: a typology to guide research and policy. Policy Internet 12(2):225–248 source

In the paper

“Human–chatbot collaboration is foreseen as an increasingly important aspect of chatbot research and applications.”

In the paper

“Chatbots may improve access to health care services (e.g. [105]) and support health-promoting behavior change (e.g. [85]), and supplement educational programs [53].”

What the sources say

• Not checked yetcites [105] Vaidyam AN, Wisniewski H, Halamka JD, Kashavan MS, Torous JB (2019) Chatbots and conversational agents in mental health: a review of the psychiatric landscape. Can J Psychiatry 64(7):456–464. https://doi.org/10.1177/0706743719828977 source
• Not checked yetcites [85] Perski O, Crane D, Beard E, Brown J (2019) Does the addition of a supportive chatbot promote user engagement with a smoking cessation app? An experimental study. Digital Health 5:2055207619880676 source
• Not checked yetcites [53] Hobert S, Berens F (2019) Small talk conversations and the long-term use of chatbots in educational settings–experiences from a field study. In: Chatbot research and design. Third international workshop CONVERSATIONS 2019. Springer, Cham, pp 260–272 source

In the paper

“There is a large and growing body of ethical and privacy knowledge to draw on, and an emerging set of guidelines and regulations on ethics and privacy for digital systems in general, and AI-based systems in particular.”

In the paper

“some studies highlighted critical aspects in using chatbots since they may sustain and even strengthen existing biases in society.”

In the paper

“However, to our knowledge, there is a lack of research on the usability, accessibility, and effectiveness of such platforms.”

In the paper

“While a range of chatbots for collaboration have been developed, there is relatively scarce research on the characteristics of collaboration with chatbots.”

In the paper

“Recent studies suggest that while chatbots may indeed serve as a low-threshold interface to information, services, and societal participation, they may also face challenges regarding bias and inclusion.”

In the paper

“recent work [58] has discovered substantial divergence in how these are written and interpreted, highlighting the complexity of designing guidelines for systems with complex social impact.”

What the sources say

• Not checked yetcites [58] McTear MF (2002) Spoken dialogue technology: enabling the conversational user interface. ACM Comput Surv (CSUR) 34(1):90–169 source

In the paper

“Chatbots are a prominent AI-based technology, and as such in principle addressed by the broader concern for ethics and privacy in technology research in general and in AI-based technology in particular.”

In the paper

“chatbot research and design are still in its relative infancy”

In the paper

“If the main characteristics of both are similar, transfer of the research results seems viable.”

In the paper

“Ethical reflections and discussion on chatbots and chatbot applications are already emerging (e.g. [65, 91]).”

What the sources say

• Not checked yetcites [65] Kretzschmar K, Tyroll H, Pavarini G, Manzini A, Singh I (2019) Can your phone be your therapist? Young people’s ethical perspectives on the use of fully automated conversational agents (chatbots) in mental health support. Biomed Inform Insights. https://doi.org/10.1177/1178222619829083 source
• Not checked yetcites [91] Ruane E, Birhane A, Ventresque A (2019) Conversational AI: social and ethical considerations. In: Proceedings for the 27th AIAI Irish conference on artificial intelligence and cognitive science—AICS 2019, pp 104–115. CEUR Workshop Proceedings, Vol-2563. http://ceur-ws.org/Vol-2563/

In the paper

“personal benefits are often not measured in monetary terms, but the main focus of personal usage is the improvement of daily life or wellbeing”

In the paper

“We anticipate that advances in the democratization of chatbots will increasingly inspire ethical discourse that ties in with higher-level discussions about chatbot applications.”

In the paper

“In business contexts such as corporate customer support scenarios, the potential impacts may be less severe, but specific corporate regulations and norms need to be considered.”

In the paper

“the current lack of coherence in the emerging field of chatbot research, which may in part be observed by the broad range of journals and conferences in which findings from chatbot research are presented, and also the lack of commonly agreed key constructs, models, and measurement instruments”

In the paper

“many research results from existing pilot studies and prototypes for specific application areas exist and may be reused as the basis for transfer and generalization”

In the paper

“chatbots are not conventional technologies, but they are developing as agents operating in social contexts”

In the paper

“the use of chatbots for social relationship, is often mainly driven by motivations for engagement and meaning-making”

In the paper

“The findings are based on discussions at the CONVERSATIONS workshop and mainly involve researchers from European organizations.”

In the paper

“future research will benefit from a greater degree of coherence and cohesiveness in the field”

In the paper

“results concerning chatbot communication, user experience, design, and technology are the basis for applying chatbots in specific application areas”

In the paper

“many research studies exist in specific domains that could possibly be transferred to others”

In the paper

“requirements in the health sector concerning privacy aspects, ethics, and trust may be significantly more demanding than similar requirements in other sectors”

In the paper

“without requiring chatbot developers to have in-depth software engineering knowledge—as exemplified in using visual programming approaches such as Blockly to chatbot development [89]”

What the sources say

• Not checked yetcites [89] Rodríguez-Gil L, García-Zubia J, Orduña P, Villar-Martinez A, López-De-Ipiña D (2019) New approach for conversational agent definition by non-programmers: a visual domain-specific language. IEEE Access 7:5262–5276 source