Strategic use of artificial intelligence (AI) is multiplying the power of data to assist companies in their hunt for oil, gas, and mining resources.
Whether by interpreting vast amounts of data or leveraging knowledge, AI is helping, but industry leaders say the workforce needs to work with the technology rather than fear it.
During a session on advances in exploration technologies for oil, gas, and mining on 23 March in Houston at CERAWeek by S&P Global, panelists emphasized the importance of data sharing, and the opportunities AI brings to the exploration sector.
Amy Callahan, Accenture’s senior managing director for US chemicals and natural resources, said AI can help pick up the slack from the lack of new people entering the industry.
“The type of people that know how to do this work have not been doing it for many years, and there’s not new talent coming into that space as much as would be necessary,” she said. “When you bring AI into that, it takes some of the need for that talent out because you get more information, you can get to your answers faster, just with less people.”
James Brady, chief digital officer for oilfield services and equipment at Baker Hughes, said AI allows a company to share the expertise of the organization over time, allowing access to knowledge of people who are already retired.
“Corporations are able to share knowledge a little bit easier than perhaps they could,” he said. “Call it more of a net human capital extender as opposed to a replacement. I like to think of it as leveraging my friends from the past.”
Hussein Shel, director and chief technologist for energy and utilities at Amazon Web Services, said traditional machine learning and AI worked really well where data was available, and now the sector is seeing progress with foundation models and domain understanding models trained on the subsurface.
“It’s starting to show some very, very promising results when it comes to the frontier in the area where we don't have a lot of data support,” he said.
Part of what sets exploration apart from other potential uses of AI is its interpretive nature, Brady noted.
“There are some rules that we learn in school. We have some knowledge about how geology is supposed to work—principles like, ‘Is the charge still there?’” he said. “The data doesn't actually tell you any of that.”
While the production side of the industry is rife with data, in exploration, “you still have to combine that intuitive, that artistic, but that deep knowledge of what's really going on,” he said.
For AI to gain traction as a solution, Brady said, it must be “explainable” during the proof-of-concept stage. “You need to be able to explain to the customer why this works. Because if you don't, [they’ll say], ‘Well, that's a one-off.’”
Drilling through a specific formation might call for automation and AI models. “It is a very problem-specific approach with AI, and then at the end, when you propose something to the client, you have to say it works—right? It's not snake oil,” Brady said.
Efforts like the Open Subsurface Data Universe (OSDU) have helped oil and gas companies communicate data more clearly, Callahan said.
“The oil and gas industry has put in place some parameters around the Open Subsurface Data Universe, things that have allowed companies to start talking between themselves, getting that data in a single format, which I think has been extremely helpful for the industry,” she said.
Using accessible data can provide value by making it possible to move from exploration to first oil faster, she said.
Brady said the industry has a way to go “with learning to share.”
Shel said one of the things that has historically hampered access to data is the way information can exist in silos. It has been estimated that people can spend more than half their time just looking for data, but such silos are dissolving, he said.
He also said partnerships between an operator, a cloud provider, and an ISP solution provider can deliver impactful projects, but the industry tends not to embrace them. There is “lots of experimentation and lots of piloting,” he said, “but unfortunately, it's very difficult for the culture to take that on.”
Callahan said newer companies, such as junior mining companies in Africa, tend to be better at adopting new methods because they “didn’t have the history” of using different technology.
When it comes to adopting AI methods, she said, the people who “need to figure out how to adopt it are like us. Everybody that is coming into the workforce, they are so passionate about all of these tools, they use them in their daily lives.”
