AI-driven mapping tools, such as satellite-based geographic information systems (GIS), often rely on historical data that may reflect social and economic biases. For example, an AI model trained on incomplete or outdated data may underrepresent informal settlements or low-income neighbourhoods, leading to their exclusion from important infrastructure planning and disaster response efforts.
A real-world case of bias is in the mapping of urban versus rural areas. AI models are trained primarily on high-quality urban data. They struggle to accurately represent rural or indigenous territories, reinforcing existing inequalities in decision making and resource allocation.
click image to enlarge
Image Source
Ethical uses of AI in geography education should ensure data sources represent diverse, accurate, and representative information to avoid misinforming students.
Inclusive mapping practices should be used to provide fair and accurate geographic insights. Find out more
Data bias and errors in GeoAI
AI systems can also perpetuate or even exacerbate biases inherent in their training data.
An example of this is where biases in mapping the data could lead for instance to the inaccurate identification of slum areas, impacting response efforts and hindering effective service distribution.
Biases in data used to train autonomous vehicles can result in unsafe situations.
Research from King’s College London revealed that widely used datasets were significantly less effective at detecting darker-skinned pedestrians compared to lighter-skinned ones, with a detection failure rate nearly eight per cent higher for darker-complexioned individuals. Find out more
These examples highlight the critical need to address AI biases to ensure safety and fairness.
Privacy and location-based data >>