ABSTRACT:
When I look at the volumes of the Journal of Management Information Systems (JMIS) of the decade or two ago and compare their contents to the present issue, I see a significant change in the individual subjects our authors have researched. I also see the constancy of purpose our information systems (IS) discipline serves. The centers of attention have changed from human-computer interface to the human-artificial intelligence (AI) collaboration, from generalized knowledge management to the generative AI (GenAI) integrating knowledge into organizational processes, from firms and supply chains to platforms. Yet the general thematic of organizational computing and organizationally-delivered information-based artifacts abides.
The momentous change is the attention devoted to the nodal aspects of human-AI relationship since the advent of GenAi. The impact of this novel form of AI as a general-purpose technology will have few precedents in its benefits, costs, and risks, and can to a degree be shaped [1]. As I previously articulated, our discipline is in a unique position to contribute to the understanding of the organizational deployment of AI. As a research collective and a scholarly field, we possess both the technological and behavioral knowledge necessary to advance the understanding of the collaboration and competition, substitution and augmentation, as the phenomena that will define our human future. We are equipped to create knowledge at the multiple levels of analysis: individual, team, community, organization, state, international, and global. Our work in IS economics should help assess the effects of GenAI and its potential successors on the labor markets and the tradeoffs that will be necessitated. The IS discipline has garnered extensive knowledge in behavioral cybersecurity that flows directly into the surpassing dangers presented by the ever more powerful AI. The deep concerns in the ethical domain and the guardrails on AI can be informed by the work published in our leading journals. Obviously, the AI-related research will be just a component of our scholarly profile, as the contents of the present JMIS issue show, but it will be a key component. AI-adjacent research, whose results are applicable or adaptable to the progress of our knowledge of AI, has been published in our journals for years, and is indeed published in this issue. We should think how to build on it in our future work on the human-AI interaction, apportionment, and outcomes.
Firms are beginning to field digital twins, that is digital replicas of specific humans in a variety of front-end operations. AI agents can play this role. Thus, for example, CVS Health deploys in its market research such digital clones of their customers, produced with the help of AI on the basis of interactions with human customers [2]. This company seeks the effectiveness of this approach as it plans producing some 100,000 such digital twins of real customers and interrogating them as it would be done in the traditional market research. Of course, digital twins are already deployed in a variety of human endeavors; AI strengthens the realism and potency of this approach. The authors of the first paper of the issue investigate the obverse use of digital twins. Lingyao (Ivy) Yuan, Antino Kim, Mike Seymour, and Alan R. Dennis study empirically the use of these avatars as agents in customer service. More specifically, the authors study the effectiveness of celebrity twins. The actions of such twins are more easily manipulated that those of their prototypes and may be, for example, personalized for the individual customers or a class of them. After all, celebrity is often a perceptual quality. The authors find celebrity twins largely effective in creating greater user engagement, overcoming algorithms aversion, and in other respects.
The study of AI aversion is offered by Kimia Ansari and Maryam Ghasemaghaei in the next paper. The infusion of AI in the organizational decision-making creates a variety of opportunities for conflict between the humans and the AI artifacts. When an AI output conflicts with the individual’s potential decision, the AI may be ignored—to the detriment of the firm—and AI aversion created. The authors study the contingencies—such as the decision criticality and the perception of the AI as a competitor versus a collaborator—in the creation of AI aversion. The authors’ results point to the necessity of the development of the organizational climate and the overall AI-human system within the firm to support the collaboration between the two. The general models of the contingencies of the agent effectiveness remain to be built—within our field, expanding the comprehensive research agenda postulated by Wessel et al. [4].
The collaboration between humans and bots (AI-based or not) is taken up in the next paper of the issue. Jiangnan Qiu, Mengjie Li, Shuangyao Gao, Rongjiang Huang, Zhongming Ma, and Yan Li present their research, focused on online knowledge communities. The work investigates the effects of the collaboration between humans and bots within these communities, deploying the data garnered from the history pages of Wikipedia entries, and based on the coordination theory. Notably, bots serve as leadoff partners in some of the articles. The human-bot trajectories yield interesting results to the authors’ analysis. The bots at present are not AI artifacts, yet the results we have here will feed into our studies of human-AI coordination patterns.
Youhyun Lee, Hsing Kenneth Cheng, and Liangfei Qiu study analytically the strategies to be adopted by firms when fielding novel deep-tech products, that is, the products deploying frontier technology, often that of AI. Should the innovating firm make its new frontier product compatible with that of an incumbent company’s? The decision to go for compatibility points to the immediately higher sales than the alternative. Yet it also favors the rival incumbent. The formal modeling in the duopolistic setting with network effects brings out the tradeoffs. It shows that the dominant strategy for the new entrant of a deep-tech product in specific circumstances may be to forgo the compatibility with a rival’s established artefact and develop an independent market for the product.
New entries to the information-technology (IT) marketplace are further the subject of the research by Franck Soh and Varun Grover. The authors approach the subject from the point of view of the numerous incumbent startups in the mobile app marketplace. The incumbents generally wish to preclude the new entries, particularly the imitators. Soh and Grover seek an answer to the question: How the incumbents can limit the potential competitors’ ability to identify the targets for the imitative entry and their willingness to imitate. The authors’ modeling deploys signaling theory to address their subject. Inter alia, they identify how social media can be deployed in the process—by both sides.
Online games are the IT attracting most direct users. There are multiple genres of these games. The authors of the next paper, Ching-I Teng, Tzu-Ling Huang, Alexander S. Dennis, Alan R. Dennis, and Gen-Yih Liao, classify them into four categories, from massively multiplayer online role-playing games (MMORPGs) to action role-playing games. Deploying the theory of action as a routine capability, the researchers proceed to surface the affordances inhering in each of these genres and those leading to the enactment of routines that enhance users’ loyalty to the game. Some of the features underlying the affordances in MMORPGs actually displease certain users as opposed to others. The success in the expansive game marketplace may depend on what the authors have identified here.
Telehealth is an important subject of study in our field. With the growth of the world’s population, the healthcare’s often precarious economics, and the need and frequent ability to equitably access remote resources, from specialists to equipment, one can confidently predict the increasing role of the research on the subject. Here, Ryan Raimi, Paul Benjamin Lowry, and Detmar W. Straub present a study of telehealth use in its psychiatric application with a view to the capability of reducing the stigma in self-disclosure. They deploy the disclosure process model to establish how the communication process can be structured to limit the effects of the perceived stigma, increase the level of trust, and lead to a fuller disclosure by the patient, and thus a more favorable course of treatment. The work complements the paper we published earlier that presents the design of an explainable AI system for mental disorder screening, which can be used as a prequel to or a part of the telehealth treatment [3].
The participation in online communities is generally dichotomized as active versus passive. A more nuanced classification of the level of activity is needed to support the participants, to raise the level of their engagement, and to increase their loyalty. Ariel D. Wigdor, Zachary J. Sheffler, and Traci J. Hess develop and validate a generalized visibility-cost framework that integrates several types of commitment and participation behaviors. As in a preceding paper, this identification can lead to an enhanced differential support for community participants and, further, to the identification and support of behaviors befitting the nature of the community.
Crowdfunding has joined the panoply of ways to finance a variety of pursuits. Charity drives are a specie of crowdfunding that appeals to the better angels of human nature. Owing to the relative dearth of such angels, these campaigns are quite limited in the degree of their success. The emotional appeals bear improvement. In the next paper, Feng Liu, Yu Chen, Yi-Chun (Chad) Ho, Subhasish Dasgupta, and Mingjie Fang apply the elaboration likelihood model and machine learning to analyze the effects of the emotional content of visual and textual appeals to charity. They find the differential effects of such appeals and present a novel model of affective modality alignment. This is an advance in our understanding of the effects of the online media and—we hope—in the success of charitable fundraising.
The concluding paper of the issue contributes significantly to our understanding of the already well-researched online reviews. Many platforms employ upvotes and downvotes to express the vox populi. Owing to the abuse of the downvotes, some platforms disable this capability. Yinan Yu, Dezhi Yin, and Warut Khern-am-nuai study the effects of that. Based on their theory-driven empirics, the researchers find well-defined effects of this platform intervention on user behavior and on the quality of reviews. Yu et al. go to the sources of this behavior—and contribute to our understanding of the phenomenon we thought we had already fully understood.