Outsourcing in the Age of Machine Learning: How Inscrutable Shared Understanding Drives Outsourcing Relationship Success

Abstract

Anchored in the IS outsourcing literature, shared understanding between the client firm and the provider has been consistently reported to support successful outsourcing engagement. A lack of shared understanding, on the other hand, is linked to deteriorating performance. In this study, we examined how shared understanding changed throughout an outsourcing engagement between a Bank (the client) and a Machine Learning (ML) service provider. Based on a longitudinal qualitative study of ML service outsourcing, from both client and provider perspectives, we found that in the early stage of the outsourcing relationship the parties bestowed shared understanding. However, in the later stages a different form of shared understanding evolved in which the ML provider intentionally limited information about how the ML model works, while the client ‘blackboxed’ the technology, while focusing on increasing the supply of data required to train the ML model and verifying its outcome. We theorize that ‘inscrutable shared understanding’ emerged in this ML outsourcing service and point out three practices pursued by the client and provider to support this new form of shared understanding.