Using algorithms to improve knowledge work

Journal article

We explore how organizations leverage algorithms to improve knowledge work in contexts where the tasks require skilled work, as distinct from routine tasks that have traditionally been the focus of academic enquiry. Drawing on a multiple-case study of four business areas in a multinational energy firm undergoing a digital transformation, we find that contrary to what the literature predicts, tasks that require skilled work can also benefit from the adoption of algorithmic solutions. To benefit, business areas engaged in two distinct pathways for transforming knowledge work. The first focuses on automating a specific task, replacing human activity with algorithms in a single task. The second involves re-engineering an entire process, whereby sequences of steps adjacent to the task at hand are redesigned on integration of an algorithm. We find that these pathways have different effects on the ability to improve knowledge work, suggesting that alignment between the task and the pathway chosen is crucial to realizing any improvement. We also find that the ability to sustain any improvement depends on the adjustment of the knowledge regime—the practices and structures that sanction knowledge. Building on these findings, we propose a general process model for the adoption of algorithmic solutions in knowledge work. In the wider context of the future of work debate, our findings challenge the prevailing notion that a task's skill requirements determine the extent to which knowledge work can be improved by algorithmic solutions.

Authors: Amaya, J; Holweg, M

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Wiley
• Journal: Journal of Operations Management
• Volume: 70
• Issue: 3
• Pages: 482-513
• Publication date: 2024-02-25
• Acceptance date: 2023-12-18
• DOI: 10.1002/joom.1296
• EISSN: 1873-1317
• ISSN: 0272-6963
• Language: English
• Keywords: knowledge work; process improvement; future of work; robotic process automation; machine learning