Images, Objects & Visual Evidence¶

March 2026

Draft

This activity guide is under development. The structure is in place but the content is not yet complete.

What this task involves¶

Many humanities disciplines work with visual and material evidence: photographs of inscriptions, pottery, coins, architectural features, artworks, maps, and archaeological finds. The task is not simply image recognition but contextual interpretation — understanding what an object is, where it fits in a typology, what its condition tells you, and how it relates to other evidence.

Where AI tools help¶

Multimodal AI models (especially Gemini and ChatGPT with vision) can describe, compare, and reason about images. Upload a photograph and ask for description, transcription of visible text, comparison with known types, or initial cataloguing. This is useful for first-pass description, OCR of inscriptions, and generating structured metadata from visual material.

What to watch out for¶

Models will describe what they see with confidence, but they lack the specialist knowledge needed for accurate identification, dating, and interpretation. They may misidentify pottery forms, misread worn inscriptions, or impose modern categories on ancient material. Visual analysis demands domain expertise that models do not have.

Worked examples¶

Coming soon.