Rethinking work in the digital age

Abstract

Incumbent firms face multifaceted challenges from considerable cost pressure and new market players. Implementing data-driven technologies such as data science, artificial intelligence (AI), and algorithmic management (AM) can be relevant measures to counteract. However, introducing data-driven technologies in the workplace leads to profound changes in various aspects of work. We state that the present understanding of the implications for work and workers within incumbent firms needs an extension in three distinct areas. First, new insights are essential on how workers adapt their work practices in response to the changing environment. Second, it remains unclear how workers build trust in AI-based systems, given concerns about the trusting belief “reliability”. Third, exploring the impact of AM on workers’ efficiency, established organizational structures, and existing work relationships within traditional organizations is crucial. A multi-method research strategy is followed to address these research gaps. By considering the complexity of our phenomenon, we set up a multiple case study to examine the reconfiguration of work practices based on the increasing usage of data science. Moreover, we apply grounded theory methodology to investigate the trust-building process of workers in AI-based systems and the utilization of an ethical framework. Finally, we measure the potential performance effects of AM in the traditional work context, based on ~12700 manufacturing errors, using linear mixed modeling. We extend this quantitative analysis by 15 confirmatory semi-structured interviews to get a deeper understanding of AM impact and find detailed explanations. The present work provides several empirical findings. First, we illustrate how engineers reconfigure the engineering role using data science work. We identify the hybrid practice of data science work, combining engineering and data science practices. Second, we develop a process for AI implementation and operation along the five phases of the AI lifecycle. We show the dynamic interplay between building trust and the acceptance level of AI reliability. Besides, the impact of an ethical framework for trust calibration is assessed. Finally, we demonstrate that AM increases workers’ efficiency and impacts existing organizational structures. In particular, human managers are essential shapers of AM and remain key supporters. Moreover, AM influences team dynamics as workers engage in collective bypassing, situational optimization, and hybrid interaction. We can make several contributions to theory and practice based on our results. We enhance research on data science work by providing evidence of the relevance of domain experts for shaping data science work and showing the specific characteristics of the new hybrid role. We extend trust theory by illustrating the dynamic process of building trust in AI, as workers are willing to reconfigure their expected level of reliability. Moreover, we highlight five supportive organizational measures and the utilization of an ethical framework. We contribute to the AM literature by validating the performance effects in traditional settings, considering socio-technical factors. Additionally, we explore the supportive role of human managers and the impact of AM on existing team dynamics. In practice, workers, leaders, and IT project team members can benefit from several recommendations for successfully implementing emerging data-driven technologies.