Geology

Podcast 2 — In conversation

Speaker A: My brother just started a degree in geology at Edinburgh — apparently the rocks in Scotland are absolutely extraordinary for studying ancient tectonic activity.

Speaker B: Scotland really is a geologist’s paradise. You have some of the oldest rocks in Europe up in the Highlands, and then the whole central belt is volcanic. Arthur’s Seat in Edinburgh is an extinct volcano, you know.

Speaker A: That is fascinating. I always thought geology was just about identifying pretty stones, but he has been explaining how it is really about understanding deep time — like, millions and millions of years compressed into a single rock face.

Speaker B: Exactly. The perspective you get from geology is completely different from any other science. When a geologist says recent, they might mean the last ice age — twenty thousand years ago. Human history is practically invisible on a geological timescale.

Speaker A: He was telling me about the difference between geology and geography — people confuse them all the time. Geography is about places and human interactions with them. Geology is specifically about the Earth materials and processes. You can have geography without rocks, but you cannot have geology without Earth.

Speaker B: That is a good way to put it. And geology is distinct from archaeology too — archaeology studies human artefacts, geology studies natural formations. Though they often work together, especially when dating sites or understanding ancient landscapes.

Speaker A: Common mistakes — well, people often use geology when they mean topography or landscape. The geology of an area is the underlying rock structure, not just what the land looks like. And you would not say I am studying the geology of this rock — you would say the petrology or just the composition. Geology is the field, not the individual specimen.

Speaker B: Right. Geology is uncountable when you are talking about the science — she studies geology — but you can use it countably for specific geological systems — the geology of the Lake District is complex. The plural geologies exists but is fairly rare, used when comparing different geological systems.

Podcast 3 — Prompt engineering

Speaker B: Geology is a powerful term for AI prompts because it introduces spatial, temporal, and material constraints that ground abstract requests in physical reality. When you mention geology in a prompt, you are signalling that you care about rock types, soil composition, terrain stability, and deep time — all of which constrain and inform design decisions.

Speaker A: So geology makes the AI think about physical foundations and material reality.

Speaker B: Precisely. Let us start with a civil engineering database:

Speaker B: PROMPT: Design a site_investigations table for civil engineering projects. Include fields for project_id, site_location, geology_classification (igneous, sedimentary, metamorphic), bedrock_depth_meters, soil_bearing_capacity_kpa, and geological_hazards (fault_lines, sinkholes, landslide_risk). Generate a report query that flags any project where the geology classification indicates karst limestone and sinkhole risk is rated high or above.

Speaker B: Geology_classification as a constrained field with specific rock types makes the AI build a schema that engineers can actually use for foundation design. The karst limestone and sinkhole risk combination is a real-world geological hazard that requires specific engineering responses.

Speaker A: That is a genuinely practical use — sinkholes are a nightmare for construction. What about a UI for geological data?

Speaker B: Here is a UI prompt:

Speaker B: PROMPT: Build a geological survey interface for a mining exploration team. The map view should overlay geology layers: bedrock type, fault lines, mineral deposits, and groundwater zones. Use distinct colour schemes for each geology layer — browns for sedimentary, greys for igneous, greens for metamorphic. Include a drill_log panel that shows geology samples at different depths with stratigraphy diagrams.

Speaker B: Geology layers, geology samples, stratigraphy diagrams — the terminology drives a domain-specific visual design. The AI produces a tool that actual geologists would recognise, rather than a generic map with random colours.

Speaker A: Colour coding by rock type is standard in geological mapping. What about a full application?

Speaker B: Here is a complete app:

Speaker B: PROMPT: Create a geothermal feasibility assessment app for homeowners. The user enters their address, and the app queries a geology database to determine the local bedrock type and thermal conductivity. Display a geology summary card showing rock type, estimated depth to stable temperature, and suitability rating for ground source heat pumps. Include a geology uncertainty warning if the local geological survey data is incomplete.

Speaker A: Geology as the data source for renewable energy decisions — that connects ancient rock science to modern sustainability. The uncertainty warning is crucial because geological surveys have coverage gaps.

Speaker B: Exactly. One more — an environmental monitoring context:

Speaker B: PROMPT: Design a groundwater monitoring dashboard for an environmental agency. Display geology cross-sections showing aquifer layers, clay barriers, and bedrock. The dashboard tracks water levels at monitoring boreholes and flags any borehole where the geology suggests potential contamination pathways between the monitored aquifer and shallower water sources. Include geology-based risk ratings for each monitoring site.

Speaker A: Geology cross-sections and contamination pathways — that is how hydrogeologists actually think about groundwater protection. The AI builds a scientifically grounded monitoring tool rather than a generic water level chart.

Speaker B: One more for a data science context:

Speaker B: PROMPT: Build a machine learning feature engineering pipeline for earthquake prediction. Extract geology features from geological survey data: rock_density, fault_line_proximity_km, geological_era, and tectonic_plate_boundary_distance. Create a geology_enriched dataset that combines these features with historical seismic records. Document which geology features showed the highest correlation with past earthquake occurrences.

Speaker A: Geology features as ML inputs — that bridges traditional earth science with modern predictive modelling. The feature documentation requirement makes the output scientifically auditable.

Speaker B: That is geology in prompts: a term that brings physical reality, temporal depth, and material constraints into any development context. Use geology deliberately and your AI produces outputs that respect the ground beneath the code.