Artificial intelligence and its performance impacts in the oil and gas industry: Challenges, insights, and evaluation approaches

摘要

Oil and gas (O&G) companies are experimenting with artificial intelligence (AI) tools in exploration, production, and trading, but few are able to provide long-term benefits. This thematic review distinguishes basic AI adoption from AI realization—the sustainable ability to embed, scale, and profit from AI through coherent data, skills, and routines. Three research strands are synthesised. First, the literature consistently identifies three driver groups—people, technical, and managerial—that accelerate or hinder realization. Second, progress is usually gauged with three measurement families: text mining of corporate filings, proxy indicators such as AI-related patents or robot density, and survey-based capability indices. Third, realised AI affects three performance domains: operational, financial, and environmental. Within O&G, value chain stage (upstream, midstream, downstream) and ownership model (state-owned and privately owned) modulate both measurement validity and outcome strength. These strands are integrated into a conceptual synthesis that links drivers to gauges and gauges to results, stressing triangulation across indicators to limit bias. Triangulated dashboards linking capability scores to these outcomes expose trade-offs, such as efficiency gains that raise energy use. Recommended actions for practitioners include: connecting operational, financial, and sustainability databases; deploying an annual “AI readiness” survey covering data, talent, and governance; and adopting shared reporting rules that enable cross-company learning. For scholars, the paper highlights measurement gaps, suggests triangulated research designs, and maps boundary conditions needing further study. Together, the insights give executives a firmer basis for allocating AI budgets and give researchers a clearer roadmap toward comparative, evidence-based evaluation of AI value in complex, asset-intensive settings. • Reviews methods to measure AI realisation and its impacts on performance. • Explores drivers of AI realisation and how they shape AI measurement gauges. • Presents a concise conceptual model linking AI drivers, gauges, and performance. • Accounts for O&G-specific factors shaping AI realisation and its performance impact. • Offers guidance for O&G firms and bodies to foster sustainable AI realisation.