This document discusses using artificial intelligence to automatically classify historical photographs. It notes that classifying a large collection of 140,000 photos would require an impractical amount of manual labor. The document then outlines challenges in classifying old photos due to uncertainty about details like location, time of day, or people present. It proposes using computer vision and deep learning techniques like OpenCV, edge detection, and TensorFlow to conduct initial classifications of photos into categories like portraits, landscapes, and architecture. The document concludes by discussing expanding the classification capabilities and identifying new applications and data sources for the techniques.