On the Ethics of Generative GeoAI: Explainability, Bias, Hallucination, Accountability, Privacy, and Trust

This peer-reviewed book chapter reviews key ethical issues in the conceptualization, development, and deployment of generative AI systems for geography and geospatial artificial intelligence (GeoAI), including model explainability, bias, hallucination, accountability, privacy, and trust.

Koutian Wu authored Section 8, “Trust in AI and GeoAI Models,” which examines:

• From probabilistic output to geospatial reality. Trust in GeoAI requires verifying that machine learning models adhere to physical and spatial laws rather than accepting statistical correlation alone. Models that predict linguistic patterns by probability can produce fluent but spatially incoherent text, a failure mode termed “geo-hallucination.”
• Balancing intelligence with physical constraints. Integrating generative models into geospatial science creates tension between algorithmic automation and human expert judgment. Treating uncertainty quantification as an ethical requirement is one way to manage this risk.
• Systemic risks and data integrity. “GeoAI collapse” describes the risk that future models are trained on the hallucinated outputs of their predecessors rather than observational data, which erodes geographic fidelity. Provenance-aware protocols help maintain trust.

Details

• Book: Volume 422, Geography According to Foundation Models
• Pages: 215-232
• Publisher: IOS Press
• DOI: 10.3233/FAIA260483

For a longer discussion, see the blog post.

Recommended citation: Mai, G., Lao, N., Zhang, J., Mao, L., Wang, Z., Wu, N., Janowicz, K., Wu, K., Rao, J., Gao, S., & Zhu, R. (2026). "On the Ethics of Generative GeoAI: Explainability, Bias, Hallucination, Accountability, Privacy, and Trust." In Geography According to Foundation Models, Vol. 422, pp. 215-232. IOS Press. DOI 10.3233/FAIA260483.
Download Paper