Innovation meets institutions: AI and the Finnish construction ecosystem

Abstract

Artificial intelligence (AI), machine learning (ML), and deep learning (DL) are technologies that have recently transformed many industries. The construction industry has traditionally been a laggard industry in terms of digital-technology adoption. When leading firms in this industry have experimented with these technologies, many of these experiments have met resistance. In this paper we take an institutional lens to study why and particular social structures appears to have contributed to the resistance and paucity of success stories. Within institutional research, we focus on research with traces to cognitive science and psychology. We have carried out a qualitative embedded multiple-case study on resistance to new technologies and how to overcome such resistance. The study involves four use cases in the Finnish construction industry: (1) automation of a material-product subcontractor’s production planning; (2) business-model innovation by contractor on how to best work across multiple construction sites at once; (3) machine learning and automation of documentation by a software firm; and (4) promotion of a vision of information sharing across organizations by the above software firm. Based on within and cross-case analyses, preliminary empirical findings are that AI, ML and DL have in the Finnish construction industry challenged institutionalized forms of organizing and workflow established long since in the industry and, until about the time of this piece of research, taken for granted. Resistance was nonetheless beginning to be overcome at the time of writing this piece of research with small-group interaction across firms – such as those in this study - - in the industry ecosystem. Human-human mediation and face-to-face encounters were building trust in and across the organizations. The implication for practice and policy is that business transformation will not quickly and autonomously transform into “impersonal” or machine-machine exchange but, before that, requires human-human mediation. “ In the long-term, AI and analytics have boundless potential use cases in E&C [i.e. engineering and construction]. Machine learning is gaining some momentum as an overarching use case (that is, one applicable to the entire construction life cycle, from preconstruction through O&M 8i.e. operations and management), particularly in reality capture (for example, in conjunction with computer vision) as well as for comparison of in situ field conditions with plans (for example, supporting twin models). Indeed, by applying machine learning to an ongoing project, schedules could be optimized to sequence tasks and hit target deadlines, and divergences from blueprints could be caught closer to real time and corrected using a variety of predetermined potential scenarios.” [1]