Artificial intelligence and analytics in practice

ABSTRACT

Artificial Intelligence and Analytics (AI&A) are used in various application areas ranging from online entertainment to healthcare. The goal of this Special Issue is to focus on how AI&A are used in practice to assist organisations to create economic value, support decision making, transforming them, and enhance employees’ skills such as communication, innovation, and decision-making. To derive actionable insights on the application of AI&A in practice, the five articles selected for this Special Issue feature both rigorous academic research and reflections and actionable lessons for professionals. This editorial contains a brief overview of the articles included in this special issue.

KEYWORDS:

• Artificial Intelligence (AI)
• analytics
• decision making
• business value

Introduction

AI and Analytics (AI&A) continue to transform organisational activities across private, public and not-for-profit sectors. Organisations are keen to harness the transformational power of AI&A to achieve game-changing results such as efficient and resilient supply chains (Cadden et al., 2021; Dennehy et al., 2021a), enhanced customer experiences (Griva et al., 2021), better decision-making (Cao et al., 2021; Duan et al., 2019; Power et al., 2019), better public services (Alshahrani et al., 2021; Vassilakopoulou et al., 2022), dynamic B2B marketing practices (Keegan et al., 2022), improved selection, prioritisation, and management of software development projects (Dennehy et al., 2021b, 2021c; Niederman, 2021; Zamani et al., 2021) and product innovation (Duan et al., 2018). Understanding how organisations use AI&A to generate actionable insights that create business and social value is critical to sustaining a competitive advantage (Berns et al., 2009; Iglesias et al., 2019; Pappas et al., 2018). At the same time, AI&A affects the way that people work, as it has both augmented and automated human activities (Johnson et al., 2022) enabling human/AI partnerships (Vassilakopoulou et al., 2022). As such, to be embedded smoothly in organisations and societies overall, re-skilling, training and several organisational and cultural changes are required (Iansiti & Lakhani, 2020; Tschang & Mezquita, 2021). Such disruptions coupled with a growing awareness of the ‘dark side’ of AI&A related to for example, inequalities, social exclusion, and fear that AI will take over the world (Davenport & Harris, 2007) call for reflections on the ethical application and governance (Mäntymäki et al., 2022a, 2022b) of AI&A so that actionable insights can be drawn for practitioners.

AI&A has created a new stream of research in retail, banking, HR, advertising, health, start-ups, fintech, manufacturing, supply chains, and so forth, and evidently, there are several scientific efforts that theoretically discuss AI&A. Topics such as how AI&A is used in organisations, what is its value, how it affects employees, how it is regulated, ethical issues around AI&A are the most prominent. However, today in academia more than ever, there is an increasing need of presenting how practitioners and organisations perceive, track and adopt AI&A in their daily services and operations (Fountaine et al., 2019). We need to hear practitioners’ voices and experiences to better understand AI&A. Practice-based frameworks and methodologies should be developed, and practitioners should share lessons learned and experiences (Delen & Zolbanin, 2018). Working with practitioners, not just for them, and making them part of the research process can lead to better solutions and generate knowledge that is even more useful for practice (Rai, 2019). For example, the creation of research champion roles at organisations (i.e. a person employed by the organisation and physically hosted at the university premises), offers a pathway to increase practice-based research and institutionalise collaborative research (Vassilakopoulou et al., 2022). Advancing clinical research (Schein, 1987, 2008) can create new opportunities for practitioners to share their experiences and insights as contributions to both research and practice. A key difference between clinical research and traditional researcher-initiated projects, such as action research or case studies (Schein, 2008, pp. 268, Figure 18.1), is that they are initiated by the organisation. Thus, practitioner-researchers become part of the research process enabling the co-production of knowledge.

To this end, this special issue contributes to practice by presenting empirical studies that showcase best practices and lessons learned in the use of AI&A in a variety of contexts such as telecom, high tech, manufacturing, etc. Our goal is to present studies that discuss how AI&A is used, what changes it causes in organisations, how it affects customer relations, how it impacts employees’ skills (e.g. innovation, decision-making), how management is affected by its adoption, what frameworks should be developed to regulate AI&A and avoid biases, and what are the lessons learned by its actual implementation in organisations.

The selected papers

We released our Call for Papers for this issue in 2020 with the aim of engaging with practitioners and researchers involved in the practical use of AI&A. All manuscripts went through a minimum of two rounds of peer review by domain experts. The Guest Editors actively engaged with the author teams throughout the review process to ensure their studies advanced understanding of the use of AI&I in practice. In total, there were 24 authors from China, France, Germany, Greece, Ireland, Italy, Kenya, South Africa, the UK and the USA, associated with accepted papers; half of them are practitioners.

We believe the papers selected for publication in this special issue ‘Artificial Intelligence and Analytics in Practice’ advance our understanding on how AI&A is used in contemporary organisations to extract value, support decision-making, and transform organisations.

Below we briefly outline the five manuscripts accepted for this Special Issue:

1. ‘Does AI control or support? Power shifts after AI system,’ by Emmanuel Monod, Raphael Lissillour, Antonia Köster and Qi Jiayin present two longitudinal case studies, the first one is from a telecom company in China, which used an AI-enabled customer service assistant exploiting Natural Language Processing (NLP) to improve customer relations. The second case concerns a small high-tech company in China that develops AI solutions for sales management via exploiting NLP, social semantics, etc. By presenting these two cases, the authors’ goal in this manuscript is to discuss the change in power relations resulting from the implementation of AIs in companies. Results indicated that after a specific period of time AI caused some unintended consequences, such as reinforcing the managers’ authority and control, reducing employees’ autonomy, and eventually leading to low acceptance of the AI system. In closing, the authors comment on the importance of symbiosis between AIs and humans.

2. ‘The impacts of artificial intelligence on managerial skills,’ by Laurent Giraud, Ali Zaher, Selena Hernandez, and Al Ariss Akram, explores this topic following a mixed-methods approach encompassing two empirical studies. The goal of the first study is to analyse interview data from 40 AI experts and identify the impact of AI on managerial skills; whereas the second aims to confirm the qualitative results via a quantitative study with 103 experts in AI. Results indicated that AI can amplify managerial skills such as communication, recruitment, complex decision-making and innovation. Also, skills such as getting information and plain decision-making are likely to be replaced, whereas imagination and leadership still cannot be replaced by AI. Findings also indicated that both technical (e.g. AI knowledge) and non-technical (e.g. judgment, collaboration, change management, risk-taking and open-mindedness) managerial skills should be cultivated to enhance the use of AI in organisations across all sectors.

3. ‘The role of management in fostering analytics: the shift from intuition to analytics-based decision-making,’ by Philipp Korherr, Dominik K. Kanbach, Sascha Kraus, and Paul Jones, focuses on how management can assist companies in shifting from intuition to data and analytics-driven decision-making. To explore this topic, the authors interviewed 22 employees from a large manufacturing company in Germany. Some departments of this company still apply traditional decision-making procedures, whereas others have shifted to analytics-based decision-making. The results indicated that six factors should be considered when companies shift to analytics-based decisions. These are: analytics infrastructure, top management and strategy, management behaviour, HR management and development, and organisation and governance. Authors also highlight that these factors should not be viewed independently, as company culture is a critical factor that affects the rest and has a major contribution to how companies can adopt analytics-driven decision support. The authors present their results describing a framework for companies to manage their transformation.

4. ‘AI ethical biases: normative and information systems development conceptual framework’ by ’Tanay Chowdhury and John Oredo takes a deep dive into the emerging need for developing frameworks to understand and tackle AI biases. In more detail, they focus on AI ethical biases for Information Systems Development (ISD). To identify the various biases, the authors conducted a narrative literature review. Their goal was to develop a conceptual framework that organises biases and categorises them under specific scopes, i.e. data, method, and implementation. Each scope is further described and developed, whereas at the same time practical examples present the identified biases and suggest possible solutions on how these can be overpassed. In closing, as an organising frame to develop this conceptual framework, the authors suggest the Cross Industry Standard Process for Data Mining (CRISP-DM), which can be viewed as one generic ISD model used in analytics projects.

5. ‘Decision support using AI: the data exploitation at telecoms in practice’ is authored by Christos-Antonios Gizelis, Konstantinos Nestorakis, Antonis Misargopoulos, Filippos Nikolopoulos-Gkamatsis, Mihalis Kefalogiannis, Poly Palaiogeorgou, Antonios M. Christonasis, Konstantinos Boletis, Theofanis Giamalis, and Chrysostomos Charisis. In this manuscript, the authors, who are current and past employees of OTE, the largest telecommunication provider in Greece, present their experience on how this large organisation implemented several AI use cases. The authors discuss the opportunities and challenges from the implementation of AI in their company during their effort to be digitally transformed and better serve their customers. These results were generated by the IT Innovation Center of the company, which possesses its own Innovation lab to run experiments and evaluate AI-enabled Proof of Concepts (PoCs) to advance the company’s services. Various PoCs are presented including chatbots, NLP, robotics, machine vision, sales forecasting and predictive maintenance. These also contribute to our understanding of how a large company identifies and implements several PoCs to develop new services using cutting-edge technologies.

Conclusion

It is evident that all the five manuscripts selected for the special issue are different in perspective, use of empirical methods, and type of AI&A studied. Taken together, however, they advance our understanding of how AI&A is used in practice, which is important for the future of IS research and practice. The selected studies provide us with significant lessons and insights on how to incorporate AI&A systems in organisations, what ethical issues might arise, which challenges emerge, what power games are at play, how we can avoid negative consequences, and, overall, how management could facilitate the adoption of such systems.

In the current special issue, some authors focused on suggesting conceptual frameworks to cope with the challenges that AI&A imposes. As such, future research can deep dive into the development of frameworks and methodologies to provide practitioners with guidelines on how to handle crucial AI&A aspects, such as ethics, biases, dark side of AI&A, black boxes, and transparency in algorithms, management behaviour, engagement practices, etc. Future research may also focus on challenges and lessons learned from various company sizes when trying to incorporate analytics and AI in their processes, and contrast and compare these with – for example, a company’s culture, maturity level, and transformation process. Specifically, a comparative analysis of the challenges faced by large companies versus start-ups would be of great interest. Further research may also focus on AI&A and control, to clearly define who is the controller and the controlee in AI systems, how this can change over the time that AI self-learns and adapt its operations, etc. Finally, while the context of the studies in this special issue included telecom, high tech, and manufacturing; future research on companies in other industries, such as retail, finance, banking and health, would be of great interest.

We hope that researchers will be motivated by this special issue and will study the use of AI and analytics in practice to further advance our knowledge in this field.

Acknowledgments

We want to thank Professor Ciara Heavin, and Professor David Sammon Editors-in-Chief of the Journal of Decision Systems, and Dr. Arif Wibisono, Managing Editor, for their continuous support and guidance throughout this process. Finally, we thank the authors for their trust and effort, and the anonymous reviewers for their time to ensure the quality of this special issue.