AI in metal detecting is not a “treasure radar”. It is a documentation assistant: read a photo, suggest what the object might be, preview restoration without risking the original. Related: restoration principles, copper and bronze, newbie mistakes.
Case 1. “What is this?” — photo ID
Problem: a worn badge/coin from the field; phone gallery loses context. In Discovera:- Create a find in the catalogue and pin the map spot.
- Upload “as found” photos (several angles, daylight).
- Run AI analysis for hypotheses and what to compare next.
Case 2. Ring restoration — pixels first
Problem: you want a “mirror clean”, but can kill patina and value. Approach:- Photograph the ring before any chemistry.
- Run AI restoration — a “how it might have looked” preview.
- Only then decide on real cleaning; if yes, follow metal-specific care and preservation.
Case 3. Place context beats a pretty frame
Problem: a nice photo without a pin and map layer is useless in a month. Keep in the same find record:- coordinates / pin;
- old-map overlay layer;
- pre-cleaning photos;
- audio/VDI note (see VDI and discrimination).
AI then gets better inputs; you get a reproducible outing archive.
When AI helps — and when it does not
| Helps | Does not replace |
|---|---|
| Type hypothesis from a photo | Lab expertise and metal assays |
| Visual “what if cleaned” | Safe chemistry protocols |
| Catalogue and outing comparison | Detecting permits and legal status |
| Draft diary description | Ethics: do not publish exact pins for valuable finds |
Ivanych — mentor, not oracle
The in-app Ivanych chat covers gear and outing scenarios in plain language. It supports learning; it is not a “dig here” guarantee. For detectors see also choosing a detector.
AI usage checklist
- Daylight photos, avoid hard glare.
- Pin + “as found” saved first.
- Read coin cost before running.
- Do not clean the original until case 2 is decided.
- AI output is a hypothesis, not a verdict.