Artificial intelligence is rapidly reshaping how geoscience is taught, communicated, and practiced. In micropaleontology, a key bottleneck remains species-level identification, an expert-driven task that is time-consuming, hard to scale, and often inaccessible to students and the wider public. Foraminifera, microscopic carbonate-producing organisms, are widely used as nature-based indicators of environmental change and for reconstructing past oceans and dating. Yet routine identification still limits the size, speed, and comparability of datasets used in biomonitoring and geological applications.
We present a project developing a human-in-the-loop, AI-assisted identification system that improves the speed, consistency, and scalability of routine foraminiferal analysis while keeping expert validation at the core. By developing an automated stereoscope-to-AI pipeline that captures high-quality photographs and immediately performs AI-based species recognition and identification.
Crucially, AI is only as good as the data behind it. Without high-quality datasets, machine learning has limited value and may amplify biases or misidentifications. We therefore aim to build an open, global, community-driven foraminifera database that compiles existing and newly generated records across taxa and regions. The platform will include standardized species “cards” (taxonomy, key diagnostic traits, and ecological/stratigraphic metadata, where available) alongside rich image collections (light microscopy and, where possible, SEM and multi-angle views). This shared resource will both support learning and training and provide robust ground truth for training, benchmarking, and generalizing AI identification models.
Together, the integrated imaging-and-identification workflow and the global reference dataset broaden access to foraminiferal expertise, strengthen teaching and guided learning, and support outreach through user-friendly interfaces that help non-specialists engage with microfossils and what they reveal about our world past, present, and future.