As artificial intelligence (AI) reduces unpleasant tasks and some of the cognitive burden of work in the oil and gas industry, the role of the human is shifting.
From delivering insights that can improve safety, increase production, and make workers’ lives easier, AI is strengthening its foothold in the industry, speakers said at the “Next-Gen AI for Energy Summit” in Houston on 3 November. And speakers predict that workers will increasingly delegate portions of their work and, eventually trust AI enough to delegate entire tasks and workflows, allowing humans to step “out of the loop” completely after proper risk assessments.
Boaz Nur, oil and gas lead at Anthropic, said during a panel session that the industry seems to have adopted generative AI more quickly than other technologies, possibly because many people are already comfortable using gen AI models in their personal lives.
And while the models are rapidly improving with each release, he said, humans should be making the decisions because “we know” what those decisions should look like.
“But in terms of all of that unpleasant work that you need to do in order to get to the point where you make a decision, I do believe that the models will be able to take that off of our hands entirely,” he added.
Brandon Croley, director of information technology and head of AI at Williams, said AI can pull together diagrams and manuals and procedures and regulations to generate plans in hours instead of days.
Working with AI in this way significantly reduces “the amount of time they spend at a desk doing that kind of work and opening up more time to go do the real value at work, the rich time that is what their real job is and makes us more valuable and more effective,” he said.
Bolo AI founder and CEO Diti Sood said that the technology is not intended to replace human judgment, particularly given the level of risk in the oil and gas industry.
“It’s really going back to that point of reducing that cognitive load that people are doing every day because so much of the work that the engineer is doing is the silent work on ‘How do I foresee, out of the 10 things I’ve seen, which one really needs my attention?’” she said. “I always tell my customers to think of it as a junior engineer because it’s not there to replace you.”
Trust and Other Challenges
It will take time for AI to earn trust, and people have differing views on how reliable technologies such as self-driving vehicles must be before they can be trusted, Nur said.
A self-driving Tesla can be “pretty scary at first. At some point you realize that if I leave the Tesla alone and don’t pay attention, it's maybe a little bit risky,” he said, adding that driving while using a cell phone and handling a large beverage can also be risky. “Humans do stupid things in cars and have accidents. But me paying attention while the Tesla drives is probably the safest of all possible vehicles on the road.”
Devon Energy Chief Information Officer Heath Satterfield said certain groups want perfection out of technologies before they can be trusted, but said failures are necessary for improving technology.
“They take a stance that perfection is the only acceptable outcome. So, a self-driving vehicle that has an accident, not hundreds of accidents or thousands of accidents, an accident is unacceptable,” he said, noting many humans have car accidents daily. “We have to have a stance and a posture that is a willingness to let things fail” so failure points can be removed from the system.
And that willingness to let things fail should start with the low-risk activities, which can then be improved before moving on to higher-risk areas, Satterfield said.
AI has earned some trust, and the expectation is that over time it will gain more.
Marcelo Bastos, senior US market intel lead at Petrobras America, noted that AI has evolved from a tool operated by an expert to an assistant helping with a task to an assistant being delegated a task.
The industry is at “the beginning of this era” of delegating tasks to AI, with autonomous systems being a next step to consider following careful risk analysis on the use case, he said.
Ways To Use
As Chevron has leaned more heavily on AI as a technology, Nuray Elci, the operator’s SVP for global marketing, services, and digital platforms, said during her keynote that Chevron’s AI-centered conversations have evolved accordingly. Instead of merely asking if the company had deployed an AI technology, she said, the focus is on making decisions faster and having a positive and measurable business impact.
She likened AI to an iceberg, in which the technology is the visible part and the hard work needed to make everything function, such as change management, governance, and process, occurs beneath the surface.
One of the most important pieces of using a technology like AI to help transform a workflow is organizational buy-in and change management, she said.
Nur, during his keynote, said workers can use AI as a thought partner that helps surface solutions they might not have considered themselves.
“Feed it all the data. Tell me what to do. Tell me what I’ve missed. Because this is hard for a human. There’s a ton of data,” he said.
As to the limitations of AI, he had this piece of advice: “Don’t let your ideas limit AI. Let AI limit your ideas.”
