Mining, Oil and Gas and Artificial Intelligence: Early Automators with Company-Reported Gains and Few Independent Measurements

Purpose. Mining and oil and gas are capital-intensive, hazardous and remote industries that were among the first to automate and are rich in data from seismic surveys and sensors. This article asks where artificial intelligence stands in them in 2026, how much of the claimed benefit can be verified, and where they may be by about 2035. Design and methods. A review that combines the share of the literature on mining, petroleum and related subjects that involves AI (OpenAlex, 2010–2025), the International Energy Agency's account of AI in oil and gas, industry analysis of autonomous haul trucks, peer-reviewed work on machine learning in geoscience, and industry surveys. Claims taken from the literature were checked against the sources cited, and company-reported results are identified as such. The scenarios are judgements built on this evidence. Findings. The share of works on these industries that mention AI rose from 1.9% in 2018 to 8.7% in 2025 (4.5 times as much, below the median across industries), with language-model works still few (233 in 2025). Automation is far advanced: an industry analysis counted about 3,800 autonomous haul trucks in July 2025, up from about 2,100 a year earlier, with China at about 2,100 and Australia at about 1,000, and the International Energy Agency reports that oil and gas companies operated 24 of the world's 500 fastest supercomputers in 2024. Peer-reviewed results are strongest in data-heavy subsurface problems, such as a deep-learning model that detected and located twice as many earthquakes as standard methods from less than a third of the stations. By contrast, the reported gains from autonomous haulage, in safety-incident frequency and productivity, come from company reports and trade press, and we found no independent study of safety, productivity and employment together. We expect autonomy to keep spreading and AI to become routine in subsurface data processing and maintenance by 2028, evidence from independent measurement to be the main gap in 2029–2032, and the pace of capital spending, which depends on commodity markets and the energy transition, to matter as much as the technology afterwards. Originality. The article separates the parts of the evidence that are peer-reviewed or measured from those that are company-reported, quantifies the growth of autonomous haulage, and shows how little independent evidence exists for a pair of industries that automate heavily.