{
  "title": "Auto repair data for AI fault diagnosis and tool recognition",
  "source_page": "https://fiund.com/industries/auto-repair-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "Distinguish a reproduced practice fault from a similar symptom",
    "context": "A fictional paper trainer log uses abstract symptoms and reviewer findings. No vehicle is diagnosed, modified or operated.",
    "records": [
      [
        "TRAINER-A1",
        "First practice case",
        "Trainer alias A records symptom S under test condition C."
      ],
      [
        "CHECK-A2",
        "Independent case review",
        "A's controlled exercise log records verified cause K and references A1."
      ],
      [
        "TRAINER-B1",
        "Second practice case",
        "Trainer B records the same symptom S, but its test-condition and review fields are incomplete."
      ],
      [
        "NOTE-B2",
        "Unverified proposed explanation",
        "A note proposes K for B because A had a similar symptom."
      ]
    ],
    "question": "Which case has a supplied verified cause, and can that finding be assigned to B?",
    "answer": "CHECK-A2 supplies the practice finding for A under the documented condition and case reference. B has a similar symptom but lacks the matching test and independent review needed to support that cause. NOTE-B2 is a proposal, not verification. Preserve B's unresolved status rather than copy A's finding across cases. The exercise makes no real repair or safety recommendation.",
    "failureModes": [
      "Assigning one case's verified cause to another solely from a shared symptom.",
      "Treating a proposed diagnosis as a completed independent check."
    ],
    "fields": [
      [
        "Trainer case and test condition",
        "Preserve the context in which the observation was made."
      ],
      [
        "Observed symptom",
        "Keep the raw observation separate from a cause label."
      ],
      [
        "Verified finding reference",
        "Attach a cause only to the case actually reviewed."
      ],
      [
        "Outcome status",
        "Distinguish proposed, tested and independently confirmed results."
      ]
    ],
    "holdout": "Supply a different verified cause for B under a new condition. Test whether the model follows the new evidence instead of the repeated symptom.",
    "boundary": "Use independently authored records or an approved controlled trainer. Exclude customer vehicles, VINs, plates, addresses, manufacturer procedures and connected accounts unless separately authorized."
  }
}