Development of a theoretical model for calibrating trust in AI to augment highways safety
Pooja Ivvala, Aya Bayramova, Chris Roberts, David J. Edwards, Iain RilliePurpose
Artificial intelligence (AI) has the potential to augment safety within the construction and highway sectors through data-driven decision-making. However, trust in AI impacts the end-users’ reliance on the technology, which directly influences safety outcomes. This research explores the relationship between the technical, organisational and human factors affecting trust in AI adoption, and the role they play in achieving safety outcomes. Premised upon the literature analysed, a theoretical model is then developed to explain the phenomenon under investigation.
Design/methodology/approach
A mixed philosophical lens was adopted, incorporating interpretivism in conjunction with inductive reasoning to conduct a systematic review of prevailing scholarly literature. The literature was extracted from Scopus (119 studies) and Web of Science (39 studies) for analysis. An inductive two-stage, mono-method systematic literature review was incorporated into a three-phase waterfall approach to develop a theoretical model that encompasses the technical, human and safety aspects of AI implementation within safety-critical environments.
Findings
Existing literature on factors affecting AI adoption and trust in AI is fragmented and human factors are viewed as a secondary determinant of the successful implementation of AI within safety-critical environments, with trust emerging as a key aspect influencing successful AI implementation. The synthesis revealed that the calibration of trust in AI is crucial to achieve safety outcomes. Therefore, a more holistic approach is needed that balances the technical, human and safety factors of AI, making human-centricity the crux of AI implementation. The theoretical model developed elucidates further upon these factors and the interactions between them.
Originality/value
This study theoretically contributes by presenting a novel theoretical model that highlights the relationship between the technical, organisational, regulatory and human factors affecting AI adoption and the role they play in achieving safety outcomes in the construction and highway sectors. Practically, this research establishes a new benchmark for industry practices, providing a robust theoretical model that can aid organisations within the construction and highway sectors to navigate appropriate AI use.