AI Resilience Report forGeothermal Production Managers

Right now, AI is mostly augmenting geothermal production managers rather than replacing them. The biggest changes are happening in the back-office and monitoring parts of the job — exactly the tasks O*NET flags as most automatable, like keeping daily logs and watching programmable logic controllers. In Indonesia, Star Energy Geothermal partnered with Kyndryl to deploy generative AI and an "AIOps" platform that uses machine learning to give real-time insights on system performance and predict equipment failures before they happen ^[1], letting operators focus on strategy instead of routine maintenance checks.

Researchers at the National Laboratory of the Rockies (formerly NREL) built a project called GOOML that creates "digital twin" models of real plants and uses machine learning on decades of well data to optimize daily processes, schedule maintenance, and detect potential trouble events ^[2]. The U.S. Department of Energy funded more than $9 million across two phases of machine-learning research, including a Phase 2 focused on advanced analytics for efficiency and automation in geothermal plant operations ^[3]. Industry leaders are also exploring how AI and high-performance computing can drive geothermal forward through digital twins and operational insight tools ^[4].

The hands-on, judgment-heavy parts — site visits, ensuring regulatory compliance, and directing crews — remain firmly human.

Adoption is moving steadily but not explosively. On the "fast" side, geothermal plants already run on sensor-rich control systems, and AI is helping the industry meet booming demand from data centers. AI is making exploration and operations quicker, cheaper, and more efficient, with startups like Zanskar using AI-native models to find and develop sites faster ^[5].

On the "slow" side, plants are safety-critical infrastructure, and Deloitte's 2025 survey found that nearly 60% of AI leaders cite integrating with legacy systems and addressing risk and compliance concerns as their top barriers to adopting agentic AI, followed closely by lack of technical expertise ^[6]. Geothermal plants must follow strict environmental and grid-reliability rules, so utilities tend to pilot AI carefully before scaling.

The hopeful takeaway: this is a growing field where AI is a teammate, not a replacement. Young people who build skills in data literacy, controls engineering, and field judgment — plus knowing how to work alongside AI tools — will be exactly the kind of managers the geothermal industry needs.