1 Introduction
2 Decision making with the help of AI
2.1 Development and current status of AI research
2.1.1 Definition and history of AI
2.1.2 AI applications
2.2 Organizational decision making
2.2.1 Decision theory and resulting challenges
2.2.2 Decision making in groups
2.3 The basic process for organizational decision making under uncertainty
3 Research methodology
Systematic literature review (Denyer and Tranfield 2009) | Content analysis approach (Mayring 2008) | Methodology in this work |
---|---|---|
1. Question formulation | 1. Material collection | 1. Question formulation |
2. Locating studies | 2. Search strategy | |
3. Study selection and evaluation | 3. Selection process | |
4. Analysis and synthesis | 2. Descriptive analysis | 4. Descriptive analysis |
3. Category selection | 5. Classification of content and interpretation | |
5. Reporting and using the results | 4. Material evaluation | 6. Result and discussion |
3.1 Search strategy
3.2 Selection process
3.3 Classification of content
-
Knowledge management with the help of AI.
-
Categorization of AI applications.
-
Impact of AI on organizational structures.
-
Challenges of using AI in strategic organizational decision making.
-
Ethical perspectives on using AI in strategic organizational decision making.
-
Impact of AI usage in strategic organizational decision making on the division of tasks between humans and machines.
Author | Title | Content | ||||
---|---|---|---|---|---|---|
Category “Knowledge management with the help of AI” | ||||||
Acharya and Choudhury | Knowledge management and organisational performance in the context of e-knowledge | The article offers an inter-organizational knowledge-sharing model to capture information which is still in employees' minds. Technology can help but structure must follow. Business strategy needs to be linked to knowledge requirements and resources allocated accordingly | ||||
Bohanec et al. (a) | Explaining machine learning models in sales predictions | The article analyzes business-to-business sales predictions based on a model that enhances team communication and reflection on implicit knowledge | ||||
Bohanec et al. (b) | Decision-making framework with double-loop learning through interpretable black-box machine-learning models | Article offers a framework for using machine learning to make sales decisions: double-loop learning | ||||
Metcalf et al. | Keeping humans in the loop: pooling knowledge through artificial swarm intelligence to improve business decision making | Demonstration of tool called Artificial Swarm Intelligence (ASI) to support group decision making and to make tacit knowledge available | ||||
Shollo and Galliers | Towards an understanding of the role of business intelligence systems in organisational knowing | Business intelligence systems balance subjectivity and objectivity, while individuals create meaning through interaction with them | ||||
Terziyan et al | Patented intelligence: Cloning human decision models for Industry 4.0 | The Pi-Mind-Methodology suggested is situated between human-only and AI-only by cloning behavior of human decision makers in specific situations: “collective intelligence as a service” | ||||
Category “Categorization of AI applications” | ||||||
Baryannis et al. (a) | Predicting supply chain risks using machine learning: The trade-off between performance and interpretability | Introduction of supply chain risk prediction framework using data-driven AI techniques and relying on the synergy between AI and supply chain experts | ||||
Baryannis et al. (b) | Supply chain risk management and artificial intelligence: state of the art and future research directions | Literature review on SC risk management finding that SC and production research rather relies on mathematical programming than on AI | ||||
Blasch et al. | Methods of AI for multimodal sensing and action for complex situations | Decisions-to-data framework with five areas of context-based AI: (1) situation modeling (data at rest), (2) measurement control (data in motion), (3) statistical algorithms (data in collect), (4) software computing (data in transit), and (5) human–machine AI (data in use) | ||||
Calatayud et al. | The self-thinking supply chain | Literature review on SC of the future proposing the self-thinking supply chain model. AI’s role in managerial decision making is still marginal as categories show | ||||
Colombo | The Holistic Risk Analysis and Modelling (HoRAM) method | HoRAM method offers scenario analysis based on AI | ||||
Flath and Stein | Towards a data science toolbox for industrial analytics applications | Offering and testing the data science toolbox for manufacturing decisions | ||||
Mühlroth and Grottke | A systematic literature review of mining weak signals and trends for corporate foresight | SLR on corporate foresight based on weak signals and changes to be detected in big data, analyzing variety of data mining techniques | ||||
Pigozzi et al. | Preferences in artificial intelligence | Survey about the presence and the use of the concept of “preferences” in artificial intelligence | ||||
Category “Impact of AI on organizational structures” | ||||||
Bienhaus and Abubaker | Procurement 4.0: factors influencing the digitisation of procurement and supply chains | Study on procurement which is claimed to be important to leverage SC collaboration. Survey with 414 participants finds that face-to-face remains more important for relationship building than AI, while AI is not expected to take over decision making completely | ||||
Butner and Ho | How the human–machine interchange will transform business operations | Study on progress of companies in implementing intelligent automation, meaning the cognitive automation to augment human intelligence. Survey by the IBM Institute for Business Value, in collaboration with Oxford Economics, with 550 technology and operations executives | ||||
Lismont et al. | Defining analytics maturity indicators: a survey approach | Descriptive survey of the application of analytics with regards to data, organization, leadership, applications, and the analysts who apply the techniques themselves | ||||
Paschen et al. | Artificial intelligence: cuilding blocks and an innovation typology | Conceptual development of typology as analytic tool for managers to evaluate AI effects. AI-enabled innovations are clustered in two dimensions: the innovations’ boundaries and their effects on organizational competencies, where the first distinguishes between product-facing and process-facing innovations and the second describes innovations as either competence enhancing or competence destroying | ||||
Tabesh et al. | Implementing big data strategies: a managerial perspective | Article provides recommendations to implement big data strategies successfully by presenting benefits and challenges, and providing real-life examples | ||||
Udell et al. | Towards a smart automated society: cognitive technologies, knowledge production, and economic growth | Survey with 2700 executives of which 38% expect AI to help them make better decisions | ||||
von Krogh | Artificial Intelligence in organizations: new opportunities for phenomenon-based theorizing | AI is an organizational phenomenon that provides two outputs: decisions and solutions (alternatives to a problem) | ||||
Category “Challenges of using AI in strategic organizational decision-making” | ||||||
Bader et al. | Algorithmic decision-making? The user interface and its role for human involvement in decisions supported by artificial intelligence | Case study on role of AI in workplace decisions demonstrating how actors can become distanced from or remain involved in decision making | ||||
Bellamy et al. | Think your Artificial Intelligence software is fair? Think again | Introducing the AI Fairness 360, an open-source toolkit for research and practitioners, based on the assumption that machine learning is always a form of statistical discrimination. It provides a platform to (1) experiment with and compare various existing bias detection and mitigation algorithms in a common framework and gain insights into their practical usage; (2) contribute and benchmark new algorithms; (3) contribute new datasets and analyze them for bias; (4) education on the important issues in bias checking and mitigation; (5) guidance on which metrics and mitigation algorithms to use; (6) tutorials and sample notebooks that demonstrate bias mitigation in different industry settings; and (7) a Python package for detecting and mitigating bias in their workflows | ||||
Canhoto and Clear | Artificial intelligence and machine learning as business tools: a framework for diagnosing value destruction potential | Framework to map AI solutions and to identify and manage the value-destruction potential of AI for businesses, which can threaten the integrity of the AI system’s inputs, processes, and outcomes | ||||
Kolbjørnsrud et al. | Partnering with AI: how organizations can win over skeptical managers | The findings of a survey with 1770 managers from 14 countries and 37 interviews with senior executives reveal that soft skills will become more important. In addition, there are different opinions on AI between mid-/ low-level managers and high-level ones. The recommendation is given that managers should take an active role and embrace AI opportunities | ||||
Lepri et al. | Fair, transparent, and accountable algorithmic decision-making processes | Analyzing the lack of fairness and introducing the Open Algortihms (OPAL) project for realizing the vision of a world where data and algorithms are used as lenses and levers in support of democracy and development | ||||
L’Heureux et al. | Machine learning with big data: Challenges and approaches | Summary of machine learning challenges according to Big Data volume, velocity, variety, or veracity | ||||
Migliore and Chinta | Demystifying the big data phenomenon for strategic leadership: Quarterly Journal | Leaders need to understand IT capabilities to make the right decisions | ||||
Singh et al. | Decision provenance: Harnessing data flow for accountable systems | Proposing data provenance methods as a technical means for increasing transparency | ||||
Watson | Preparing for the cognitive generation of decision support | Interviews with 11 experts on how AI will affect organizational decision-making lead to 10 steps of how to prepare | ||||
Whittle et al. | Smart manufacturing technologies: Data-driven algorithms in production planning, sustainable value creation, and operational performance improvement | Survey with 4400 participants finds AI to be supporting cooperation and multi-stakeholder decision making | ||||
Category “Ethical perspectives on using AI in strategic organizational decision-making” | ||||||
Bogosian | Implementation of moral uncertainty in intelligent machines | Presenting computational framework for implementing moral reasoning in artificial moral agents | ||||
Cervantes et al. | Autonomous agents and ethical decision-making | Presentation of computational model of ethical decision making for autonomous agents, taking into account the agent’s preferences, good and bad past experiences, ethical rules, and current emotional state. The model is based on neuroscience, psychology, artificial intelligence, and cognitive informatics and attempts to emulate neural mechanisms of the human brain involved in ethical decision making | ||||
Etzioni and Etzioni | AI assisted ethics | To answer question of how to ensure that AI will not engage in unethical conduct, article suggests a oversight programs, that will monitor, audit, and hold operational AI programs accountable: the ethics bot | ||||
Giubilini and Savulescu | The artificial moral advisor. The “ideal observer” meets artificial intelligence | Introducing the “artificial moral advisor” (AMA) to improve human moral decision making, by taking into account principles and values and implementing the positive functions of intuitions and emotions in human morality without their downsides, such as biases and prejudices | ||||
Hertz and Wiese | Good advice is beyond all price, but what if it comes from a machine? | Experiment with 68 undergraduate students to explore whether humans distrust machine advisers in general, finding that they prefer machine to human agents on analytical tasks and human to machine agents on social tasks | ||||
Kirchkamp and Strobel | Sharing responsibility with a machine | There is a difference between human/human and human/AI teams: people behave more selfish when being part of a group, but not in case of being in groups with AI | ||||
Neubert and Montañez (2019) | Virtue as a framework for the design and use of artificial intelligence | Overview of how google is using AI (negatively) incl. Googles overall goals for AI. Introduction of virtue dimensions for decision making assigned to AI and defined for AI | ||||
Parisi | Critical computation: Digital automata and general artificial thinking | The article focuses on transformation of logical thinking by and with machines | ||||
Shank et al. | When are artificial intelligence versus human agents faulted for wrongdoing? Moral attributions after individual and joint decisions | A survey with 453 participants on human, AI, and joint decision making reveals that AI always is perceived as less morally responsible than humans | ||||
Vamplew et al. | Human-aligned artificial intelligence is a multiobjective problem | The Multiobjective Maximum Expected Utility paradigm leads to human-aligned intelligent agents. Goals are important for focusing AI decisions and limit consequences | ||||
Webb et al. | “It would be pretty immoral to choose a random algorithm” | Presentation of “UnBias” project which tries to implement fairness and transparency in algorithms: Survey with case studies on limited resource allocation problems asking 39 participants to assign algorithms to scenarios | ||||
Wong | Democratizing algorithmic fairness | Analyzing the political dimension of algorithmic fairness and offering a a deliberative approach based on the accountability for reasonableness framework (AFR) | ||||
Category “Impact of AI usage in strategic organizational decision-making on the division of tasks between humans and machines” | ||||||
Agrawal et al. | Exploring the impact of artificial intelligence: prediction versus judgment | Machines can learn judgment from humans over time | ||||
Anderson | Business strategy and firm location decisions: testing traditional and modern methods | Four models of decision making are analyzed, finding that “human intelligence still rules” | ||||
Bolton et al. | The power of human–machine collaboration: Artificial Intelligence, Business Automation, and the Smart Economy | Analyzing data from several databases to make estimates regarding the impact of artificial intelligence (AI) on industry growth, how AI could change the job market, reasons given by global companies for AI adoption, and leading advantages of AI for international organizations | ||||
Jarrahi | Artificial intelligence and the future of work: human-AI symbiosis in organizational decision making | AI can assist humans in predictive analytics, gathering and interpreting data, and should augment, not replace, human decision makers. In higher levels, visionary thinking is more important than data for decisions | ||||
Klumpp and Zijm | Logistics innovation and social sustainability: how to prevent an artificial divide in human–computer interaction | Article provides theoretical framework, describing different levels of acceptance and trust as a key element of human–machine relationship for technology innovation, mentioning the danger of an artificial divide. Based upon the findings of four benchmark cases, a classification of the roles of human employees in adopting innovations is developed | ||||
Lyons et al. | Certifiable trust in autonomous systems: making the intractable tangible | AI systems need to be tested appropriately to promote trust among users | ||||
Parry et al. | Rise of the machines: a critical consideration of automated leadership decision making in organizations | The authors model a scenario where AI substitutes humans in decision making, claiming that high safeguarding is needed. AI systems tend to overweigh objective criteria over subjective ones. AI can only assist to find a vision for leadership teams but not take the decisions alone at this high level | ||||
Rezaei et al. | IoT-based framework for performance measurement: a real-time supply chain decision alignment | Article offers a SCOR-based decision-alignment framework for SC performance management with human intelligence-based processes for high-level decisions and machine-based ones for operational decisions, both linked by machine intelligence | ||||
Schneider and Leyer | Me or information technology? adoption of artificial intelligence in the delegation of personal strategic decisions | A survey with 310 participants on willingness to delegate strategic decisions reveals that low situational awareness enables delegation to AI, which implies the same risk as not delegating parts of the decision to AI due to too much self-confidence in a situation | ||||
Shrestha et al. | Organizational decision-making structures in the age of artificial intelligence | Comparison of human and AI-based decision making along five dimensions: specificity of the decision search space, interpretability of the decision-making process and outcome, size of the alternative set, decision-making speed, and replicability. Based on this, offering of framework for combining both for organizational decision making (full human to AI delegation; hybrid-human-to-AI and AI-to-human sequential decision making; and aggregated human–AI decision making) | ||||
Smith | Idealizations of uncertainty, and lessons from Artificial Intelligence | AI is more adequate for prescriptive than descriptive decision making. In decisions under uncertainty, psychological context needs to be valued | ||||
Yablonsky | Multidimensional data-driven artificial intelligence innovation | Analysis of relationship between Big Data, AI and Advanced Analytics to define AI innovation from a managerial perspective and not a technical or architectural one. Development of multidimensional AI innovation taxonomy framework, that can be used with a focus on data-driven human–machine relationships, and applying AI at different levels of maturity |
4 Results and discussion
4.1 Distribution of articles per year, journal, and research methodology
4.2 Using AI as support for strategic organizational decision making
4.2.1 Knowledge management with the help of AI
4.2.2 Categorization of AI applications
Application | Top–down/bottom–up | Use case | Useful for step | Sources | ||||
---|---|---|---|---|---|---|---|---|
Artificial neural networks | Bottom–up | Optimization (e.g., supplier selection, SCM processes) Prediction (e.g., production losses or trends) | 3, 4, 5 | |||||
Bayesian Networks | Bottom–up | Probability assessment Impact assessment of specific outcome Detect connections/create networks | (3), 4, 5 | |||||
Decision trees | Rather top–down | classification If–then rules Detect connections | 3, 4 | |||||
Fuzzy systems | Top–down and bottom–up | if–then rules Prediction based on high variety of data Optimization (e.g., SCM processes) Preference definition (e.g., recommendation systems) | 3, 4 | |||||
k-means | Top–down | Clustering (e.g., for customer segmentation) Classification | 3, 4 | |||||
Nearest neighbour | Top–down | clustering Preference detection, if no utility function exists | 3, 5 | Pigozzi et al. (2016) | ||||
Pattern mining (incl. association-rule mining, business mining) | Top–down and bottom–up | Classification (e.g., for recommendation systems) | 2, 3, (4) | |||||
Regression | Rather top–down | Probability assessment Classification Detect connections (e.g., for sales or customer behavior forecast) | 3, 4, 5 | |||||
Support vector modelling (SVM) | Bottom–up | Classification Generalization Detect connections Inclusion of preference function, if available | 3, 4, 5 | |||||
Special applications provided by the sample | ||||||||
Datascience toolbox | Top–down and bottom–up | Predictive information implemented in business processes (e.g., manufacturing systems); human interpretation needed | 2, 3, 4 | Flath and Stein (2018) | ||||
Holistic risk application method (HoRAM) | Bottom–up | dynamic method Combination of consequences with probability of occurrence Simulation-based scenario approach | 3, 4, 5 | Colombo (2019) | ||||
Self-thinking supply chain | Top–down | continuous performance monitoring High connectivity between physical and digital systems | 2, 3, 4 | Calatayud et al. (2019) |