Proceedings of the 54th Hawaii International Conference on System Sciences | 2021

Artificial Intelligence-based Assistants (AIA)

Rainer Schmidt Rainer Alt Alfred Zimmermann Munich University oAS Leipzig University Reutlingen University [email protected] [email protected] alfred.zimmermann@ reutlingen-university.de

providers (e.g. Alexa, , ) as well as numer- 1. Introduction ous chatbots that created with the open-source devel- opment kits. For example, the Alexa universe already (AI) has received much at- comprises 47’000 applications, which are referred to tention due to the recent progress in several technolog- as “skills” [6]. This is also linked to a growth of appli- ical areas such as image detection, translation, and de- cation fields and interaction modes, which explains cision support [1]. Established businesses and many why assistant platforms are also recognized as general- start-up businesses are eagerly discussing how they purpose technologies [7]. Following the understanding can gain a competitive advantage from complement- that information systems are socio-technical in nature, ing their products, services and processes with AI. In AI-based assistants should be framed as a step toward fact, based on the research in the AI domain since sev- humanizing technology and work environments in the eral decades, a broad variety of promising application digital economy. This opens the stage for many re- fields were suggested where AI might add business search questions that shall be addressed in the mini value. Meanwhile, applications are not limited to sim- track. ple structured problems, but even applications higher complexities are feasible, which require higher levels 2. Goal of Mini Track of “intelligence”. To avoid discussions on the ambiv- alent notion of “intelligence”, it shall refer to tasks in- The mini track “Artificial Intelligence-based As- volving perception, processing, action and learning sistants” was initiated in 2021 for the first time to pro- [2]. Many applications are possible along these activi- mote the scientific exchange on AI-based assistants. It ties, in particular a user’s interaction via natural lan- should enable researchers to present and discuss inno- guage. vative approaches, methodologies, models, processes, AI-based assistants employ technologies such as etc. to design, implement, deploy, operate and opti- natural language processing, predictive analytics, ma- mize such assistants for the digital economy. The AI- chine learning, as well as voice recognition and gener- based assistants mini track sought submissions to the ation. They unlock business value through automating following (non-exhaustive) topics: processes, intensifying customer interaction, reducing • Applications of AI-based assistants in the digital errors, and speeding up interactions to name a few [3]. economy In general, two main impacts may be observed [4]. • Methods and models to design, develop, imple- First, virtual personal assistants and chatbots, such as ment, deploy, manage and monitor AI-based as- Amazon Alexa and Google Home, enable the interac- sistants. tion of human beings with machines by voice, replac- • Methods, tools, and approaches to capture user ing standard human-computer interaction via mouse, behavior using techniques such as process min- keyboard, and screen of an application. Second, AI- ing to derive recommendations for actions. based assistants replace human beings interfacing two • New business models and processes based on AI- application systems. These kinds of AI-based assis- based assistants. tants are also subsumed under the term Robotic Pro- cess Automation (RPA) [5], handling interactions pre- 3. Accepted Papers viously performed by humans to pair two application systems. Fourteen papers were submitted to the mini track The success of AI-based assistants may be de- with seven of them being accepted after a rigorous re- scribed with the various offerings from the big tech view process with two phases.

URI: https://hdl.handle.net/10125/71106 978-0-9981331-4-0 Page 4021 (CC BY-NC-ND 4.0) Marco Gärtler and Benedikt Schmidt investigate In their paper “All About the Name: Assigning the practical challenges of building a virtual assistant Demographically Appropriate Names to Data-Driven in their paper “Practical Challenges of Virtual Assis- Entities”, Soon-gyo Jung, Joni Salminen, and Bernard tants and Voice Interfaces in Industrial Applications”. J. Jansen develop a method for assigning demograph- They use a representative and simplified case from of ically appropriate names to data-driven entities. The knowledge retrieval domain to identify two significant value of this method is removing the time-consuming obstacles: First, user acceptance is lower than for com- human effort parable GUI-based systems and, second, a dispropor- Communication with conversational agents (CA) tional amount of effort to get all details and having a has become increasingly important. Therefore robust system. Rangina Ahmad, Dominik Siemon and Susanne In their paper “A for Assistant Robra-Bissantz examine, whether individuals rate Platforms”, Rainer Schmidt, Rainer Alt, and Alfred communication satisfaction of a CA similar to their Zimmermann develop the notion of assistant platforms own personality as higher. The results of their experi- and elaborate a conceptual model that supports busi- ment indicate that highly extraverted CAs are gener- nesses in developing appropriate strategies. The model ally better received in terms of social presence and consists of three building blocks, an architecture that communication satisfaction. Further, the author found depicts the components as well as the possible layers that incorporating personality into CAs increases per- of an assistant platform, the mechanism that deter- ceived humanness. mines the value creation on assistant platforms, and the ecosystem with its network effects, which emerge 4. Acknowledgements from the multi-sided nature of assistant platforms. Its main purpose is to advance the understanding of assis- We wish to thank all authors who submitted pa- tant platforms and to trigger future research. pers to the Artificial Intelligence-based Assistants Due to the recent proliferation of the concept of mini track for having shared their work with us, the anthropomorphism, human likeness in technology, has (virtual) participants creating fruitful discussion, as increasingly attracted researchers’ attention. To create well as the members of the AI-based Assistants Pro- a comprehensive understanding, Mengjun Li and Ay- gram Committee, who made a remarkable effort in re- oung Suh reviewed empirical studies, in their paper viewing the submissions. We also thank the organizers “Machinelike or Humanlike? A Literature Review of of HICSS 2021 for their help in organizing this mini Anthropomorphism in AI-Enabled Technology”. track, which was conducted fully virtually for the first Based on their analysis, they discuss potential research time. gaps and offer directions for future research Using voice assistants for shopping incorporates 5. References elements of risk affecting when and how they are con- [1] McAfee, A., and E. Brynjolfsson, Machine, Plat- sidered trusted relationship partners. Therefore, Alex form, Crowd: Harnessing Our Digital Future, W. Mari and René Algesheimer investigate in their paper W. Norton & Company, New York, 2017. “The Role of Trusting Beliefs in Voice Assistants dur- [2] Dietzmann, C., and R. Alt, “Assessing the Business ing Voice Shopping the effect of trusting beliefs to- Impact of Artificial Intelligence”, Proceedings wards voice assistants on decision satisfaction through 53. Hawaii International Conference on System the indirect effect of consideration set size in the con- Sciences, (2020). text of voice shopping. Their findings show a positive [3] Zimmermann, A., R. Schmidt, and K. Sandkuhl, direct effect of trust on customer’s satisfaction and a “Strategic Challenges for Platform-based Intelli- mediating role of set size, confirming consumers bias gent Assistants”, Proceedings KES-Conference, towards default choices. Elsevier Science Inc. (2020). Neda Mesbah, Christoph Tauchert, Peter Bux- [4] López, G., L. Quesada, and L.A. Guerrero, “Alexa mann examine in their paper “Whose Advice Counts vs. Siri vs. Cortana vs. Google Assistant: a com- More – Man or Machine? An Experimental Investiga- parison of speech-based natural user interfaces”, tion of AI-based Advice Utilization” differences in the International Conference on Applied Human advice utilization depending on whether it is given by Factors and Ergonomics, Springer (2017), 241– an AI-based or human advisor. Drawing on task-tech- 250. nology the authors investigated the relationship be- [5] van der Aalst, W.M.P., M. Bichler, and A. Heinzl, tween task, advisor and advice utilization. The find- “Robotic Process Automation”, Business & Infor- ings show that compared to human advisors, judges mation Systems Engineering 60(4), 2018, pp. utilize advices of AI-based advisors more when the ad- 269–272. vice is similar to their own estimation.

Page 4022 [6] Major, D.J., D.Y. Huang, M. Chetty, and N. Feam- ster, “Alexa, Who Am I Speaking To? Under- standing Users’ Ability to Identify Third-Party Apps on Amazon Alexa”, arXiv preprint arXiv:1910.14112, 2019. [7] Helpman, E., General purpose technologies and economic growth, MIT press, 1998.

Page 4023