By Elisabeth Scibiorski, Senior Geo-Data Scientist
Machine learning has transitioned from curiosity to standard practice in economic geology. SEG 2025 showcased both successful ML applications and important discussions about data quality, best practices, and limitations. Key trends include maturing AI adoption, increased focus on risk management, and growing interest in real-time geometallurgical workflows.
The Society of Economic Geologists (SEG) conference brought together economic geologists from industry, government and academia at the Brisbane Convention & Exhibition Centre from 26-29 September, 2025. While the event focused on the most recent, impactful developments in economic geology, my main takeaway was that machine learning (ML) is no longer a curiosity in economic geology, but is becoming standard practice.
Machine learning’s growing presence
Machine learning appeared throughout the conference program, not just in dedicated sessions but also in several talks on topics ranging from deposit-scale geochemical and geometallurgical studies, to continental-scale prospectivity modelling. The widespread adoption seems to reflect the industry’s growing familiarity with the basic concepts of data science, and an increased understanding of the ways in which ML can be utilised.
The “Machine Learning and AI” session where I presented exemplified this evolution. Some talks optimistically showcased successful data science applications across geoscientific problems, while others cautioned about the limitations of ML, particularly when misapplied by inexperienced practitioners (the aphorism ‘a little knowledge is a dangerous thing’ comes to mind), or when poor quality data produces models that fail outside training environments. These discussions around data quality, data science best practices, and the limitations of ML are familiar themes for us at Datarock. And yes, the session competed directly with the AFL Grand Final kickoff! Yet the room was surprisingly well attended (and not only by international visitors), proving that even the AFL can’t keep geologists away from good data science.
My presentation was about some work we’ve done in the last couple of years to develop a new workflow for the automated stratigraphic modelling of an iron ore deposit in the Pilbara. The talk was well received, generating positive feedback and detailed technical questions afterward.
Datarock’s contribution
Datarock had a strong presence throughout SEG. Our Principal Geometallurgist, Wendy Ware, spoke on data fusion in the pre-conference workshop “From Orebody Knowledge to Geometallurgy: Discovery to Closure”. My colleague in the Applied Science team, Senior Geo-Data Scientist, Katie Silversides, presented some recent work on automating coal-seam detection using MWD (measure-while-drilling) data. At our booth, Tim (our Business Development Manager) and Wendy were kept busy by a steady stream of former, current and prospective clients – as well as a few students – interested in finding out more about Datarock, or just stopping by to say hello and have a chat.
Three takeaways and trends: maturing ML, risk, and geometallurgy
Three main trends emerged from the conference. First, the adoption of machine learning is accelerating as industry recognises these tools’ value within economic geology. Both Wendy and our CEO, Liam Webb, noted that the conversation around AI and data analytics continues to mature, though the industry is still defining how AI and geoscience co-exist in routine workflows.
Second, risk management was a major topic of conversation, with companies examining spending decisions and the timing of resource allocation. Possibly related, there were also more discussions of geotechnical considerations than expected.
Finally, there seemed to be significant interest in what Datarock can do within the geometallurgical field. While I’m not a geometallurgist myself, Wendy feels the field is shifting from simply collecting data towards being able to integrate the insights from these data in real-time, which presents an opportunity to embed data-driven workflows that enable better decision-making.
Overall, the conference demonstrated that economic geology is embracing the digital transformation thoughtfully, balancing enthusiasm with critical evaluation of methods and results.
