When a computer speaks institutional talk: Exploring challenges and potentials of virtual assistants in face-to-face advisory services

Abstract

Advisory services are a highly sensitive form of collaboration: they rely on a clear distribution of roles between human participants who act according to an implicit set of practices and scripts. As such, they do not offer a specific role to a virtual assistant. At the same time, the technological improvements make the promise that institutional settings may be soon complemented with technology that allows for asking questions using natural speech, understands the context, and provides answers based on online processing of data. This article explores challenges and potentials of virtual assistants in advisory services while analyzing data from interviews and a workshop with clients and advisors from financial advisory services. It links the insights from the field with the institutional talk perspective. The findings unveil, that the concerns and hopes of potential users relate to their position and an implicit understanding of what an advisory service is about. This calls for careful and attentive design approach towards virtual assistants in advisory services.

Collaborating with technology-based autonomous agents

“…First, the panel discussion highlighted how the design of autonomous technology-based agents is intertwined with the tasks they are intended to support. It is reasonable to assume that technology-based agents will increasingly assist teams with small changes to common collaboration practices, such as recording, transcribing, and archiving meetings in a knowledge repository (Dolata et al , 2019; Hoffman, 2016). These systems can collect work products communicated through e-mails and chat messages (Hoffman, 2016) and collect, analyze, synthesize, predict, and identify patterns by utilizing their computational power (Gil et al , 2014; Ransbotham et al , 2017).…”

Supporting Online Customer Feedback Management with Automatic Review Response Generation

“…Nevertheless, we claim it is worth questioning the concealment practice. Given that many users can sense that machines are at work in a service context [21,22,24], honesty and transparency about who does what might be valued more than tactics following the Matryoshka principle. We call for intensifying research that tests how users react to disclosure.…”

“…Information Systems (IS) research has recognized the potential of applying NLP-based bots-as-a-service [21,22,23]. The research explores the use of such technologies in face-to-face settings [24,25], synchronous chat [26,27], as well as asynchronous communication like answering emails [24].…”

Vero: A Method for Remotely Studying Human-AI Collaboration

“…In the human-AI collaboration literature, some of the most pressing open questions involve user perceptions of AI humanness, capabilities, and transparency [7,4,3]. Researchers within IS have started to investigate such questions, with recent work examining topics ranging from symbiotic co-evolution of human-AI teams [8], trust of intelligent systems [9,10] and interaction design [11,6,12,13] to developing scales for measuring perceived AI intelligence and anthropomorphism [5].…”

“…Methodologies Existing human-AI collaboration research has been somewhat limited by the need to develop a physical or virtual agent for participants to interact with, meaning that some existing work has relied on methods that may not generalize to real team contexts, such as workshops and interviews [11].…”