During a June 10 special session at the Unconventional Resources Technology Conference (URTeC) in Houston, panelists discussed the latest advancements in Generative AI and machine learning for unconventionals.
AI—not just a ‘no brainer’ solution. Session attendees first heard from Nuny Rincones, Reservoir Engineering Innovation Manager at ConocoPhillips. For the past three years, Rincones said, she has led a team focused on reservoir analytics, probabilistic forecasting and machine learning tactics.
“Embracing technology just to embrace technology is not the answer,” Rincones cautioned. While she lauded the benefits of AI applications in the field, she also stressed the importance of truly understanding workflows, and applying AI where it will have the most impact.
“When you have upwards of 40,000 wells in the Permian basin, this is where tools like cloud computing can help,” said Rincones. “We are using AI everywhere.”
Machine learning methods. Next in the panel of speakers was Uchenna Odi, PhD., a Petroleum Engineering Specialist, AI Team Lead, and Digital Transformation Team Lead for Aramco Americas.
Odi gave a detailed presentation of his experience in studying and using automated machine learning for facies prediction scenarios.
Facies prediction involves determining the rock type (or facies) present at a specific location using data from well logs and other geophysical measurements. This process is crucial for understanding subsurface and reservoir properties, and optimizing hydrocarbon E&P.
“During a project early in my career, we were trying to better understand drilling through facies,” explained Odi.
When comparing traditional approaches versus machine learning, Odi realized significant time savings could be achieved.
“Traditional methods for facies prediction and characterization took months,” Odi said. “Traditional machine learning can cut that time down to a matter of weeks, and automated machine learning can reduce study time to just days.”
Five years ago, Odi said, he set out to learn more about what kind of startups were coming out of Silicon Valley, and how they might be applicable in the oil and gas realm. He has since implemented Automated Machine Learning (AutoML), using the software DataRobot to develop AI algorithms for data modeling.
From the lab to the field. A third panelist, Yitian Xiao, has explored similar innovations. Currently with the Deep-Time Digital Earth International Research Center, Xiao previously spent 25 years with ExxonMobil as a data scientist, and several years with SINOPEC’s Petroleum Exploration & Production Research Institute.
In 2025, Xiao said, he shifted to a research lab to work exclusively with machine learning models and GenAI. “The energy industry has tremendous potential to utilize advanced AI tools, said Xiao. “This can fundamentally change the way we analyze data.”
Xiao is a Senior Consultant for the GeoGPT Project being developed at Zhejiang Lab in China. GeoGPT is a non-profit, open-source Large Language Model (LLM) for use in geosciences. While still in prototype mode, the team hopes GeoGPT will be publicly available later this year.
Unlocking shale and tight oil with AI. To close out the session, Travis Clark, a data scientist with Chevron, touched on several tools Chevron is working on to maximize value in the oilfield. “We have quite a bit of data from the Permian,” Clark said.
Chevron is one of the Permian’s top producers, and among the many oil majors who are integrating AI into its operations. With lower oil prices creating a tougher-to-meet breakeven price on projects, efficiency is more crucial than ever.
Clark presented several tools being developed by Chevron, including Lateral Inflow, which works to improve well production and frac design, and SimOps Schedule Optimizer, which “predicts and minimizes potential frac hits,” Clark said.
