Uploaded May 2026 | Updated September 2026, 2 weeks ago
Australia’s geoscience datasets—seismic, wells, core imagery, geochemistry, remote sensing, maps, reports and monitoring time series—are among our most valuable sovereign assets. Yet much of the AI value chain is being shaped offshore, using models trained on data we don’t control, can’t audit, and that may not reflect Australian conditions, values, or regulatory expectations.
In this public talk, Dr Sue Keay will set the strategic case for Australian-made (sovereign) AI in geoscience: not to “compete with the Silicon Seven,” but to make targeted national investments that keep capability, talent, and accountability onshore. She will connect the discussion to the ATSE framing on national AI opportunity, emphasising practical pathways to turn sovereign datasets into productivity, resilience, and industry impact, while strengthening transparency, provenance, and ethical use of data.
Professor Flora Salim will then show what “building it here” looks like technically. She will outline how multimodal and time-series AI, including self-supervised learning and foundation-model approaches can learn from heterogeneous geoscience signals (imagery + text + sensor streams + spatial/temporal context). Using examples from her research in sensor fusion, spatio-temporal modelling, and robust AI for real-world systems, Flora will discuss the opportunity to create domain-adapted models that better capture Australian environments and workflows, supporting faster interpretation, improved decision support, and new capabilities across operational observability, logistics, exploration, hazard monitoring, environmental stewardship, and infrastructure planning.
The session will close with a practical discussion on what geoscience organisations can do now: data readiness, governance and provenance, partnership models, and how to identify high-impact “first deployments” that earn trust and deliver measurable value.
0:00 - Introduction
5:05 - Seminar
Australia’s geoscience datasets—seismic, wells, core imagery, geochemistry, remote sensing, maps, reports and monitoring time series—are among our most valuable sovereign assets. Yet much of the AI value chain is being shaped offshore, using models trained on data we don’t control, can’t audit, and that may not reflect Australian conditions, values, or regulatory expectations.
In this public talk, Dr Sue Keay will set the strategic case for Australian-made (sovereign) AI in geoscience: not to “compete with the Silicon Seven,” but to make targeted national investments that keep capability, talent, and accountability onshore. She will connect the discussion to the ATSE framing on national AI opportunity, emphasising practical pathways to turn sovereign datasets into productivity, resilience, and industry impact, while strengthening transparency, provenance, and ethical use of data.
Professor Flora Salim will then show what “building it here” looks like technically. She will outline how multimodal and time-series AI, including self-supervised learning and foundation-model approaches can learn from heterogeneous geoscience signals (imagery + text + sensor streams + spatial/temporal context). Using examples from her research in sensor fusion, spatio-temporal modelling, and robust AI for real-world systems, Flora will discuss the opportunity to create domain-adapted models that better capture Australian environments and workflows, supporting faster interpretation, improved decision support, and new capabilities across operational observability, logistics, exploration, hazard monitoring, environmental stewardship, and infrastructure planning.
The session will close with a practical discussion on what geoscience organisations can do now: data readiness, governance and provenance, partnership models, and how to identify high-impact “first deployments” that earn trust and deliver measurable value.
0:00 - Introduction
5:05 - Seminar
