Geography According to ChatGPT -- How Generative AI Represents and Reasons about Geography
Krzysztof Janowicz, Gengchen Mai, Rui Zhu, et al.
This paper examines how ChatGPT and similar AI systems understand and represent geography, rather than just measuring accuracy. The researchers explore three key concerns: whether AI models have brittle defaults that change with small wording tweaks, how biases can emerge from combining multiple tasks, and whether focusing only on factual recall misses deeper geographic understanding.
language modelsfoundation modelsgeographic reasoningbias and fairness