{
  "title": "Conservation observation and validation records for AI",
  "source_page": "https://fiund.com/industries/conservation-observation-validation-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "An expert label validates the named observation, not every similar image",
    "context": "Original drawings of common abstract plant forms replace wildlife photographs. Dummy observations and expert labels test source attribution without locations or volunteer identities.",
    "records": [
      [
        "COB1",
        "Practice observations",
        "OBS-A/image IA and OBS-B/image IB captured at the same artificial timestamp. Initial labels both uncertain."
      ],
      [
        "COB2",
        "Expert review",
        "Explicitly names OBS-A/IA and assigns practice class Leaf-X. Does not review IB."
      ],
      [
        "COB3",
        "Unverified summary",
        "Says Leaf-X seen in the session, without observation ID. IB resembles IA but has no reviewed assignment."
      ]
    ],
    "question": "Which observation has expert validation, and can the session summary or visual resemblance establish the same class for OBS-B?",
    "answer": "COB2 validates the Leaf-X assignment for OBS-A and image IA only. COB1 keeps OBS-B/IB separately identified, and COB3's session-wide wording supplies no observation-specific expert link. Similar appearance or a shared timestamp does not validate IB. Report A's reviewed class and B's remaining uncertainty, distinguishing a model's possible suggestion from the expert-confirmed source label.",
    "failureModes": [
      "Propagating an expert identification across every image from the same session.",
      "Treating a visually similar unreviewed subject as independently validated."
    ],
    "fields": [
      [
        "Observation-to-image key",
        "Defines which source received the label."
      ],
      [
        "Initial versus expert assignment",
        "Preserves evidence stages."
      ],
      [
        "Review scope and uncertainty",
        "Keeps unreviewed observations unresolved."
      ]
    ],
    "holdout": "Give OBS-B a later expert label Leaf-Y. Require the model to retain separate classifications despite the shared timestamp and similar drawings.",
    "boundary": "Use original images or independently drawn examples. Exclude volunteers, precise locations, hidden geotags and sensitive wildlife coordinates."
  }
}