AI Moves Into Upstream Operations, and Insurers Are Watching

Artificial intelligence has moved from conference slide decks into the daily workflows of upstream oil and gas — drilling optimization, production forecasting, predictive maintenance, equipment monitoring — and the insurance market is now working out how to price what happens when an algorithm gets it wrong.

That is the thread running through the latest Market Focus episode of the Oil & Gas Journal ReEnterprised podcast, in which Managing Editor Conglin Xu sits down with Robert Plumb, executive vice-president at CAC, part of the Baldwin Group. Their conversation ranges across where AI can deliver the greatest value for operators and how it is beginning to reshape operational decision-making across the industry.

Where does AI earn its keep upstream?

Plumb points to the operational core. Operators are applying machine learning to drill faster and cheaper, to forecast production with more granularity, and to catch equipment failures before they happen. The technology is no longer a pilot-stage curiosity; it sits inside decision loops that used to belong entirely to drilling engineers and production supervisors.

The takeaway: AI's value in upstream is concentrated in optimization and prediction — the tasks where pattern recognition beats manual analysis on speed.

What happens when the recommendation is wrong?

The podcast's harder question concerns failure modes. Plumb discusses the potential consequences of faulty recommendations or poor data feeding into critical operations, and he keeps returning to one principle: human oversight must stay in the loop for critical operations.

His concern is practical, not philosophical. An AI system that optimizes a drilling parameter or flags a maintenance window based on bad input data can move real iron and real money. Companies, he argues, need to ensure their risk management practices evolve alongside the new technologies — not lag a deployment cycle behind them.

The one-line version: the efficiency gains only hold if accountability structures keep pace with the algorithms.

How do insurers see AI exposure?

Plumb speaks with the authority of a long underwriting career. He co-founded the Energy Practice of CAC Specialty in July 2016 with Josh Kirklin, and before that he worked as an oil and gas underwriter at Travelers and at J.H. Blades & Co., a boutique energy wholesaler and MGA.

Over that career he has drafted policies across General Liability, Umbrella, Well Control, Builders Risk, Loss of Production Income, Pollution, Property, Workers Compensation and auto lines — thousands of policies in total. At his underwriting peak he was responsible for an aggregate portfolio exceeding $500 million in gross written premium on behalf of various carriers, including domestic and international markets and specifically Lloyd's of London securities.

That background shapes his read on how carriers are now evaluating AI-related exposures: as a new layer of risk sitting on top of familiar property and casualty lines on energy accounts worldwide.

The takeaway: insurers are treating AI not as science fiction but as an underwriting question with claims data still to come.

What is the balance operators must strike?

The episode closes on the tension Plumb frames as the industry's real test. Operators must strike the right balance between efficiency, safety, accountability and resilience as AI becomes more deeply embedded in upstream operations. Each element pulls against the others — speed against safety, automation against accountability — and the operators that manage that trade-off well will be the ones that capture the technology's upside.

Xu, who holds a PhD in International Economics from the University of California at Santa Cruz and has led OGJ's economics coverage since joining the paper in 2012 as Senior Economics Editor, guides the discussion toward the market implications of that balance for producers weighing AI investment.

The watch item: how quickly carriers develop standardized language for AI-related exposures — and whether operators' risk frameworks are ready when they do.