Macrotrends to Watch Out for in the AI Exploration Space
Explore macrotrends in AI for mineral exploration, from multimodal models and foundation models to uncertainty-driven exploration and AI-powered discovery.
All
AI 69
Geology 26
Benchmarking 24
Machine Learning 16
Evaluation 12
Agentic AI 10
Open Source 9
Reinforcement Learning 5
China 4
Complexity Theory 3
AI
Geological Map Processing Suite
Geological map processing for AI agents: Stratigraphic Amenity turns Microsoft PEACE into a Python SDK and MCP server for layout, georeferencing and overlays.
How Reports From The Future Draws Its Star Maps
UMAP and HDBSCAN turn arXiv papers and AI discourse into star maps: how Reports From The Future embeds, projects and clusters documents into constellations.
Groundtruth Dynamic Benchmarking (Geology Edition) v1.0.0 Is Live
Groundtruth Dynamic Benchmarking v1.0.0: a pipeline that builds benchmark questions and rubrics for any corpus, with a runnable geology example.
Ground Truth: Dynamic Benchmarking in Highly Specific Fields
Explore dynamic benchmarking in highly specific fields to measure knowledge acquisition, evaluate fine-tuned models, and benchmark models on custom corpora.
Dynamic Benchmarking: How We Did It
Fine-tuned model evaluation that separates learned geology from memorised training data: how we built the benchmark, and the five problems we hit.
Meet the Geocluster Research Harness
The Geocluster Research Harness, open-sourced: a browser IDE with a geology AI agent and 50+ geoscience tools, built on Cline and a local MCP server.
Survival is the Only Reward: The Observer Effect in AI Benchmarking
How benchmarking a self-improving agent breaks down when its reward is an emergent property of a changing world: four assumptions that stopped holding.
Feralisation
Feralisation: why AI harmlessness training may not last once continuous learning exposes models to Darwinian selection that turns domestic animals feral.
Survival is the Only Reward
An open-ended AI feedback loop: an LLM survives by holding disk space in Linux containers and retrains on what worked. Sliding-window retraining won.