190-3 What I Learned Teaching Geoscience With AI

Abstract:

Putting generative AI into a geoscience course is easy. Designing a course in which AI contributes without taking over the intellectual work is considerably harder.

Teaching a variety of asynchronous, online, introductory geoscience courses, I have developed AI-Guided Scientific Inquiries that treat generative AI as a bounded learning and thought partner rather than an answer generator or authority. The experience has changed how I think about AI-supported geoscience education.

Several practical lessons have emerged. Giving students access to AI is not the same as teaching them to use it well. Better prompting is useful, but it is not the educational goal. Students need practice questioning responses, requesting evidence, correcting errors, recognizing unsupported assumptions, working through uncertainty, and deciding when another source should remain authoritative. AI can also be too helpful: immediate, polished explanations can remove productive struggle unless an activity deliberately returns scientific reasoning to the student.

Geoscience provides particularly useful conditions for this approach. Students can use AI to explore interacting variables, compare competing explanations, test scenarios, examine cascading Earth-system effects, and reconsider an initial interpretation while still remaining responsible for evidence and conclusions.

The design has also revealed important friction. AI can become verbose, overconfident, agreeable, or disconnected from assignment context. Students may experience repeated documentation as burdensome even when they value the underlying inquiry. Some resist required AI use for academic, environmental, economic, or philosophical reasons. Those reactions do not necessarily indicate failure. A skeptical student who learns to question, verify, constrain, and reject AI may be demonstrating exactly the kind of judgment the course is intended to develop.

My central lesson is that successful AI integration is not about maximizing AI use. It is about designing the relationship between student, instructor, scientific evidence, and AI so that the technology can participate without becoming the authority. The most important question is not what AI can do for a geoscience student, but what intellectual work the student must continue to do.

Geological Society of America Abstracts with Programs. Vol. 58, No. 5, 2026

doi: 10.1130/abs/2026AM-22459

© Copyright 2026 The Geological Society of America (GSA), all rights reserved.