How AI-Based Systems Can Induce Reflections: The Case of AI-Augmented Diagnostic Work

2023 | Management Information Systems Quarterly | Citations: 0

Authors: Abdel-Karim, Benjamin; Pfeuffer, Nicolas; Carl, K. Valerie; Hinz, Oliver

Abstract: This paper addresses a thus-far neglected dimension in human-artificial intelli ... Expand

Abstract: This paper addresses a thus-far neglected dimension in human-artificial intelligence (AI) augmentation: machine-induced reflections. By establishing a grounded theoretical-informed model of machine-induced reflection, we contribute to the ongoing discussion in information systems (IS) regarding AI and research on reflection theories. In our multistage study, physicians used a machine learning-based (ML) clinical decision support system (CDSS) to see if and how this interaction can stimulate reflective practice in the context of an X-ray diagnosis task. By analyzing verbal protocols, performance metrics, and survey data, we developed an integrative theoretical foundation to explain how ML-based systems can help stimulate reflective practice. Individuals engage in more critical or shallower modes depending on whether they perceive a conflict or agreement with these CDSS systems, which in turn leads to different levels of reflection depth. By uncovering the process of machine-induced reflections, we offer IS research a different perspective on how such AI-based systems can help individuals become more reflective, and consequently more effective, professionals. This perspective stands in stark contrast to the traditional, efficiency-focused view of MLbased decision support systems and also enriches theories on human-AI augmentation. Collapse

Topics: artificial intelligence decision making decision support system algorithmic transparency recommender system

Methods: cognitive walkthrough survey theory development qualitative interview Student's t-test

Theories: learning theory information systems theory theory of abduction

Teaching Analytics in Colleges of Business

2023 | HICSS | Citations: 0

Authors: Gramling, Regina; Rainer, Kelly

Abstract: Business Analytics has become a common term in all businesses. The importance of ... Expand

Abstract: Business Analytics has become a common term in all businesses. The importance of analytics in the modern business environment has made it necessary to include analytics as a foundational course for all business majors. There is much debate on what content should be taught in analytics courses and what is the role of computational skills versus the skills needed to interpret and communicate the results that are needed as part of the analytics process? These problems point to the essential difference between analytics and statistics. As we discuss in this paper, analytics is a process that employs statistical tools. We first discuss the business analytics process which we feel should be taught in all business analytics courses. We then propose a two-course sequence, Business Analytics I and Business Analytics II, to be taught as required courses in colleges of business. Collapse

Topics: analytical information system data management decision support system marketing management advertising management

Methods: machine learning statistical hypothesis test linear regression analysis multiple regression logistic regression

Control enactment in context: Understanding the interaction of controlee and controller perceptions in inter-organisational project teams

2023 | Information Systems Journal | Citations: 0

Authors: McCarthy, Stephen; O'Raghallaigh, Paidi; Li, Yuzhu; Adam, Frédéric

Abstract: Control is necessary for aligning the actions of management (i.e., controllers) ... Expand

Abstract: Control is necessary for aligning the actions of management (i.e., controllers) and subordinates (i.e., controlees) around common goals. The enactment of control often fails in practice; however, as controlee perceptions may not match those of controllers, leading to a myriad of possible outcomes. Through an interpretive case study of two inter-organisational IT projects, we reveal how controlees' appraisals and responses to controls are context-dependent and play out across multiple levels (e.g., personal, professional, project and organisational contexts). We build on a coping perspective of IS controls to theorise the ‘coping strategies’ that controlees pursued relevant to these contexts and the ‘coping routes’ followed when combining different consecutive coping strategies. We find the process need not end with the selection of a single strategy but can potentially continue as both the controller and controlees make ongoing readjustments. While Behavioural Control Theory traditionally assumes the presence of a single control hierarchy, interorganisational IT projects are multi-level entities that amalgamate different structures and cultures. Our study moves beyond the existing assumptions of Behavioural Control Theory to discuss how a controller's choice of activities shapes the salience of different contexts in controlee appraisals. Collapse

Topics: IT project interorganizational system organizational context electronic health systems development

Methods: case study theory development conceptual modelling qualitative interview theoretical contribution

Theories: control theory

Promoting Learning Through Explainable Artificial Intelligence: An Experimental Study in Radiology

2023 | International Conference on Information Systems | Citations: 0

Authors: Ellenrieder, Sara; Kallina, Emma; Pumplun, Luisa; Gawlitza, Joshua Felix; Ziegelmayer, Sebastian; Buxmann, Peter

Abstract: The deployment of machine learning (ML)-based decision support systems (DSSs) i ... Expand

Abstract: The deployment of machine learning (ML)-based decision support systems (DSSs) in high-risk environments such as radiology is increasing. Despite having achieved high decision accuracy, they are prone to errors. Thus, they are primarily used to assist radiologists in their decision making. However, collaborative decision making poses risks to the decision maker, e.g. automation bias and long-term performance degradation. To address these issues, we propose combining findings of the research streams of explainable artificial intelligence and education to promote human learning through interaction with ML-based DSSs. We provided radiologists with explainable vs nonexplainable decision support that was high-vs low-performing in a between-subject experimental study to support manual segmentation of 690 brain tumor scans. Our results show that explainable ML-based DSSs improved human learning outcomes and prevented false learning triggered by incorrect decision support. In fact, radiologists were able to learn from errors made by the low-performing explainable ML-based DSS. Collapse

Topics: decision support system decision support explainable artificial intelligence decision making artificial intelligence

Methods: experiment qualitative interview magnetic resonance imaging parametric test ANOVA

Knowledge management mechanisms and common knowledge impacts on the value of knowledge at individual and organizational levels

2023 | International Journal of Information Management | Citations: 0

Authors: Sabherwal, Rajiv; Steelman, Zach; Becerra-Fernandez, Irma

Abstract: Knowledge management (KM) impacts have typically been examined on the value of k ... Expand

Abstract: Knowledge management (KM) impacts have typically been examined on the value of knowledge to organizations but not individuals. This paper uses survey data (186 employees) and qualitative data (300 + employees) at the National Aeronautics and Space Administration to examine the effects of four KM mechanisms (social learning, learning from others through IT, learning from IT, and IT-based knowledge substitution) and common knowledge on the individual and organizational value of knowledge. The results indicate that common knowledge and the two interpersonal mechanisms (social learning, learning from others through IT) enhance the perceived value of knowledge at both levels, but both mechanisms that rely on IT for the knowledge (learning from IT, IT-based knowledge substitution) reduce the perceived knowledge value at one level and have no significant effect at the other level. Further, common knowledge strengthens the positive effects of both interpersonal mechanisms, and weakens the negative effects of IT-based knowledge substitution, on perceived value of knowledge at both levels. Collapse

Topics: organizational value explicit knowledge IS impact knowledge sharing service desk

Methods: qualitative interview chi squared test focus group structural equation modeling design artifact

Theories: assimilation theory of learning

Showcase: A Data-Driven Dashboard for Federal Criminal Sentencing

2023 | Journal of the Association for Information Systems | Citations: 0

Authors: Vo, Ace; Plachkinova, Miloslava

Abstract: The main purpose of the Sentencing Reform Act of 1984 was to provide more unifo ... Expand

Abstract: The main purpose of the Sentencing Reform Act of 1984 was to provide more uniformity in sentencing and reduce interjudge disparity. Subsequently, the act created the federal sentencing guidelines to offer judges a possible sentencing range for offenses. However, since these recommendations were based on historical data, the guidelines amplified existing biases and increased inequality and the disproportionate sentencing of minorities. To address this problem, we developed an artifact called "ShowCase"-a data-driven dashboard-that is grounded in penal theory, organizational context theory, social bonds theory, and triangulation notion in design theory. The artifact helps judges make fairer and more objective decisions by integrating a variety of data points. We used a design science research methodology and mixed methods to guide the development and evaluation of the proposed dashboard. Our research inquiry revealed the legal and extralegal factors that contribute to more equitable judicial decisions. We also found support for integrating data science and more diverse viewpoints in the sentencing process. Our study shows that a validated data-driven dashboard can be used to promote fairness, objectivity, and transparency in the criminal justice system. Collapse

Topics: digital dashboard organizational context decision making criminality societal evaluation criteria

Methods: design artifact survey triangulation literature study design process

Theories: information systems theory

Understanding information systems success: a hybrid view

2022 | European Journal of Information Systems | Citations: 0

Authors: Jewer, Jennifer; Compeau, Deborah R

Abstract: The information systems (IS) success model, introduced in 1992, provided IS rese ... Expand

Abstract: The information systems (IS) success model, introduced in 1992, provided IS research with a comprehensive set of dependent variables for project success. While the model addresses both process and variance considerations, the latter has dominated the research. Concurrently, the benefits of hybrid theories have been discussed in the literature, and there have been calls for an integrated view of IS success that includes the process perspective. We build on this momentum by presenting a hybrid model based on a longitudinal case study of the development and implementation of a patient-flow decision-support system at a large not-for-profit hospital. Our model remains true to the DeLone and McLean framing but elaborates on the process elements. The hybrid model expands our ability to analyse multiple dimensions of IS success and integrates diverse research findings into the IS success model, providing a revised version for future research to extend. Collapse

Topics: IS success information system use digital dashboard user satisfaction human computer interaction

Methods: qualitative interview case study survey conceptual modelling focus group

Theories: information systems success model complex adaptive system

Applying and Extending the Theory of Effective Use in a Business Intelligence Context

2022 | Management Information Systems Quarterly | Citations: 6

Authors: Trieu, Van-Hau; Burton-Jones, Andrew; Green, Peter; Cockcroft, Sophie

Abstract: The benefits that organizations accrue from information systems depend on how ef ... Expand

Abstract: The benefits that organizations accrue from information systems depend on how effectively the systems are used. Yet despite the importance of knowing what it takes to use information systems effectively, little theory on the topic exists. One recent and largely untested exception is the theory of effective use (TEU). We report on a contextualization, extension, and test of TEU in the business intelligence (BI) context, a context of considerable importance in which researchers have called for such studies. We used a mixed methods, three-phase approach involving instrument development (n = 218), a two-wave cross-sectional survey (n = 437), and three sets of follow-up interviews (n = 33). The paper contributes by (1) showing how TEU can be contextualized, operationalized, and extended, (2) demonstrating that many of TEU’s predictions hold in the BI context while also revealing ways to improve the theory, and (3) offering practical insights that executives can draw on to improve the use of BI in their organizations. Collapse

Topics: data integration decision making system quality data quality decision support system

Methods: qualitative interview survey mixed method literature sample data transformation

Theories: representation theory theory of effective use

THE LIFE CYCLE OF DATA LABELS IN ORGANIZATIONAL LEARNING: A CASE STUDY OF THE AUTOMOTIVE INDUSTRY

2022 | European Conference On Information Systems | Citations: 0

Authors: Eirich, Joscha; Fischer-Pressler, Diana

Abstract: Data labels are an integral input to develop machine learning (ML) models. In c ... Expand

Abstract: Data labels are an integral input to develop machine learning (ML) models. In complex domains, labels represent the externalized product of complex knowledge. While prior research discussed labels typically as input of ML models, we explore their role in organizational learning (OL). Based on a case study of a German car manufacturer, we contextualize a framework of OL to the use of labels in organizations informing about organizational members who work with labels, requirements of labelbased tools, label-related tasks, and impediments of label-related task performance. From our findings, we derive propositions about the role of labels in OL and outline future research opportunities. Our results inform theory about the role of labels in OL and can guide practitioners leveraging labels to create and transfer knowledge within organizations. Collapse

Topics: task productivity explicit knowledge organization learning knowledge sharing intrinsic motivation

Methods: case study qualitative interview random forest classification decision tree classification business process modeling

Theories: sociotechnical systems theory

Data-driven decision-making in credit risk management: The information value of analyst reports

2022 | Decision Support Systems | Citations: 0

Authors: Roeder, Jan; Palmer, Matthias; Muntermann, Jan

Abstract: Other than banks, non-financial companies also continuously monitor and analyze ... Expand

Abstract: Other than banks, non-financial companies also continuously monitor and analyze their credit risk exposure to avoid possible counterparty defaults. Credit default swaps are commonly used financial instruments that provide information on a counterparty’s creditworthiness. Although this metric can provide crucial insights, the un­ derlying price dynamics often remain unknown and require further explanation. Data-driven decision-making is a key concept for identifying these reasons and supporting and justifying decisions. In this paper, we provide such justifications by applying sentiment and topic analysis to company-related financial analyst reports. While the contents of financial news have been analyzed in the past, analyst reports can offer additional insights, as seasoned analysts use them to disseminate in-depth research to experienced investors. This analysis examines 3386 analyst reports covering constituents of the Dow Jones Industrial Average Index in the period from 2009 to 2020. The results suggest that even when established credit risk indicators and financial news are considered, the sentiment and a subset of topics are correlated with changes in the credit default swap spread, indicating a fundamental relationship between quantitative risk metric and analyst reports. We find that analyst reports contain information related to the change in credit default swap spreads, an insight that helps to improve our understanding of existing risk assessments. The outcome indicates that banks or corporate risk managers can benefit from complementing established financial metrics and even financial news data with new insights derived from analyst reports. Collapse

Topics: accounting risk management decision making mergers and acquisition Python

Methods: regression analysis method topic model machine learning longitudinal research regression diagnostic