The UK government released an AI-generated map of Britain's peatlands claiming 95% accuracy, but local residents and farmers quickly identified widespread inaccuracies including misidentifying rocks as peat and woods as degraded peatland.
The Department for Environment, Food and Rural Affairs (Defra) released an AI-generated peat map with claims of 95% accuracy to help combat peat erosion, reduce flood risk and prioritize funding for restoration. The map was designed to identify all areas of peatland across England and warned that 80% of England's peatlands were in degraded condition. However, when local residents examined their areas, they found comprehensive errors including Cat Frampton's farm on Dartmoor where rocky soil was identified as peat, and Haytor with massive rock outcrops was mapped as having 87cm of degraded peat. Tim Ashton in Shropshire found that Shakespeare's Soulton bluebell wood was incorrectly mapped as prime peat, while actual degraded peat fields were unlisted. The map showed supposed peat deposits along dry stone walls, hedges and trees. According to researcher Lily-belle Sweet, the problems arose because the AI lacked sufficient real data points about peat to learn accurately from, and there were insufficient real-life checks to identify and correct mistakes. The National Farmers Union reported receiving significant feedback about problems from members across multiple counties including Dorset and the Lake District.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed