Overcoming Resistance to Change: Artificial Intelligence in the Energy Sector

Abstract

Background and Objectives: Artificial Intelligence (AI) promises productivity, safety, and sustainability gains in asset-intensive sectors; however, outcomes in the energy sector remain uneven. The sector's safety-critical operations, capital intensity, and stringent regulatory requirements make it a particularly demanding context for AI adoption, where technical performance alone is insufficient to ensure value. This study treats adoption as a socio-technical process rather than a tooling decision. It addresses three research questions: (RQ1) how workforce development and change leadership shape acceptance and sustained use; (RQ2) which organisational and governance conditions mitigate resistance and enable legitimate deployment; and (RQ3) under what conditions adoption yields operational reliability and environmental performance aligned with decarbonisation goals. Methodology: A qualitative, multi-case design triangulated a semi-structured interview with a senior manager, Likert-scale surveys of mid-level managers and technical staff, and analysis of internal policies and strategy documents. Data were anonymised, thematically coded using a blended inductive–deductive approach, organised in a shared codebook, and synthesised across cases to map convergences and divergences in readiness, workforce development, and governance. Intercoder reliability was assessed, and disagreements were resolved through adjudication and iterative refinement of the codebook across cases. Triangulation maintained a transparent chain of evidence. Ethical safeguards included obtaining informed consent, maintaining confidentiality, and obtaining prior approval from the relevant institutional authorities. Main Results: Three reinforcing levers shape adoption outcomes. First, broad-based capability building beyond specialist teams prevents benefits from concentrating in expert enclaves and reduces brittle scale. Second, communicative governance that couples transparency with contestability, through model cards, bias tests, validation reports, and explicit appeal rights, earns trust, curbs shadow workarounds, and improves safety culture. Third, a tight workflow fit that minimises cognitive overhead at the decision point accelerates legitimate use and strengthens links to emissions monitoring and predictive-maintenance outcomes. Thin training coverage fosters anxiety about substitution and slows diffusion; structured upskilling and precise recourse mechanisms are associated with higher confidence, productivity, and clearer sustainability pathways. Discussions: Algorithmic accuracy alone does not determine value; legitimacy and uptake hinge on people's and process readiness. The three levers translate literature on dynamic capabilities, AI readiness, and human responses to automation into operational guidance: invest in non-specialist literacy, institutionalise assurance and recourse, and engineer for workflow ergonomics in safety-critical contexts. Environmental gains materialise where oversight intensity, data quality, and targeted use cases align, indicating that governance quality conditions the conversion of adoption into credible emissions reductions. A responsible scale is pragmatic: build organisation-wide competence, communicate for legitimacy, and design for workflow fit. Conclusions: Leaders should fund training coverage and design rather than headline hours, equip non-specialists to interpret model outputs, pair performance artefacts with participatory routines, and treat explainability as a usability requirement. Policymakers can reinforce these conditions by shifting from technology-neutral principles to auditable process standards that couple AI investment with reskilling and data-quality obligations. Future research should extend the design longitudinally and incorporate behavioural metrics to test causal links. The contribution is a field-tested playbook linking human capability, assurance, and workflow design to durable, auditable value in safety-critical energy contexts.

Ähnliche Arbeiten