I ran vision AI over 27,142 rural homes to score visible distress. The most useful result was about the homes I couldn't see.

Original research applying vision AI to detect physical distress in rural residential properties across 8 Southeastern counties. The post emphasizes counterintuitive findings and, notably, the insights gained from analyzing cases where the AI failed to produce assessments. This represents practical machine learning application with honest discussion of limitations.