Eliciting Latent Knowledge (ELK)

{
  "id": "eliciting-latent-knowledge",
  "numericId": null,
  "path": "/knowledge-base/responses/eliciting-latent-knowledge/",
  "filePath": "knowledge-base/responses/eliciting-latent-knowledge.mdx",
  "title": "Eliciting Latent Knowledge (ELK)",
  "quality": 91,
  "readerImportance": 24,
  "researchImportance": 37,
  "tacticalValue": null,
  "contentFormat": "article",
  "tractability": null,
  "neglectedness": null,
  "uncertainty": null,
  "causalLevel": null,
  "lastUpdated": "2026-03-13",
  "dateCreated": "2026-02-15",
  "llmSummary": "Comprehensive analysis of the Eliciting Latent Knowledge problem with quantified research metrics: ARC's prize contest received 197 proposals, awarded \\$274K, but \\$50K and \\$100K prizes remain unclaimed. CCS achieves 4% above zero-shot on 10 datasets; Quirky LMs recover 89% of truth-untruth gap with 0.95 anomaly detection AUROC. Research investment estimated at \\$1-5M/year across ARC (3 permanent researchers), EleutherAI, and academic groups. Problem fundamentally unsolved—no proposal survives ARC's builder-breaker methodology.",
  "description": "ELK is the unsolved problem of extracting an AI's true beliefs rather than human-approved outputs. ARC's 2022 prize contest received 197 proposals and awarded \\$274K, but the \\$50K and \\$100K solution prizes remain unclaimed. Best empirical results achieve 75-89% AUROC on controlled benchmarks (Quirky LMs), while CCS provides 4% above zero-shot. The problem remains fundamentally unsolved after 3+ years of focused research.",
  "ratings": {
    "novelty": 5,
    "rigor": 7,
    "actionability": 5,
    "completeness": 7.5
  },
  "category": "responses",
  "subcategory": "alignment-theoretical",
  "clusters": [
    "ai-safety"
  ],
  "metrics": {
    "wordCount": 2494,
    "tableCount": 24,
    "diagramCount": 3,
    "internalLinks": 8,
    "externalLinks": 25,
    "footnoteCount": 0,
    "bulletRatio": 0.03,
    "sectionCount": 33,
    "hasOverview": true,
    "structuralScore": 15
  },
  "suggestedQuality": 100,
  "updateFrequency": 90,
  "evergreen": true,
  "wordCount": 2494,
  "unconvertedLinks": [
    {
      "text": "Alignment Research Center's (ARC) 2021 report",
      "url": "https://www.alignment.org/blog/arcs-first-technical-report-eliciting-latent-knowledge/",
      "resourceId": "5efa917a52b443a1",
      "resourceTitle": "ARC's first technical report: Eliciting Latent Knowledge"
    },
    {
      "text": "Alignment Research Center",
      "url": "https://www.alignment.org/",
      "resourceId": "0562f8c207d8b63f",
      "resourceTitle": "alignment.org"
    },
    {
      "text": "Zou et al. (2023)",
      "url": "https://arxiv.org/abs/2310.01405",
      "resourceId": "5d708a72c3af8ad9",
      "resourceTitle": "Representation Engineering: A Top-Down Approach to AI Transparency"
    },
    {
      "text": "alignment.org",
      "url": "https://www.alignment.org/blog/arcs-first-technical-report-eliciting-latent-knowledge/",
      "resourceId": "5efa917a52b443a1",
      "resourceTitle": "ARC's first technical report: Eliciting Latent Knowledge"
    },
    {
      "text": "arXiv",
      "url": "https://arxiv.org/abs/2310.01405",
      "resourceId": "5d708a72c3af8ad9",
      "resourceTitle": "Representation Engineering: A Top-Down Approach to AI Transparency"
    },
    {
      "text": "Alignment Research Center",
      "url": "https://www.alignment.org/",
      "resourceId": "0562f8c207d8b63f",
      "resourceTitle": "alignment.org"
    },
    {
      "text": "ARC ELK Report",
      "url": "https://www.alignment.org/blog/arcs-first-technical-report-eliciting-latent-knowledge/",
      "resourceId": "5efa917a52b443a1",
      "resourceTitle": "ARC's first technical report: Eliciting Latent Knowledge"
    },
    {
      "text": "LessWrong ELK Discussion",
      "url": "https://www.lesswrong.com/posts/qHCDysDnvhteW7kRd/arc-s-first-technical-report-eliciting-latent-knowledge",
      "resourceId": "37f4871113caa2ab",
      "resourceTitle": "LessWrong"
    }
  ],
  "unconvertedLinkCount": 8,
  "convertedLinkCount": 0,
  "backlinkCount": 1,
  "hallucinationRisk": {
    "level": "low",
    "score": 25,
    "factors": [
      "no-citations",
      "high-rigor",
      "conceptual-content",
      "high-quality"
    ]
  },
  "entityType": "approach",
  "redundancy": {
    "maxSimilarity": 15,
    "similarPages": [
      {
        "id": "probing",
        "title": "Probing / Linear Probes",
        "path": "/knowledge-base/responses/probing/",
        "similarity": 15
      },
      {
        "id": "sleeper-agent-detection",
        "title": "Sleeper Agent Detection",
        "path": "/knowledge-base/responses/sleeper-agent-detection/",
        "similarity": 15
      },
      {
        "id": "scheming-detection",
        "title": "Scheming & Deception Detection",
        "path": "/knowledge-base/responses/scheming-detection/",
        "similarity": 14
      },
      {
        "id": "arc",
        "title": "ARC (Alignment Research Center)",
        "path": "/knowledge-base/organizations/arc/",
        "similarity": 13
      },
      {
        "id": "debate",
        "title": "AI Safety via Debate",
        "path": "/knowledge-base/responses/debate/",
        "similarity": 13
      }
    ]
  },
  "coverage": {
    "passing": 7,
    "total": 13,
    "targets": {
      "tables": 10,
      "diagrams": 1,
      "internalLinks": 20,
      "externalLinks": 12,
      "footnotes": 7,
      "references": 7
    },
    "actuals": {
      "tables": 24,
      "diagrams": 3,
      "internalLinks": 8,
      "externalLinks": 25,
      "footnotes": 0,
      "references": 4,
      "quotesWithQuotes": 0,
      "quotesTotal": 0,
      "accuracyChecked": 0,
      "accuracyTotal": 0
    },
    "items": {
      "llmSummary": "green",
      "schedule": "green",
      "entity": "green",
      "editHistory": "red",
      "overview": "green",
      "tables": "green",
      "diagrams": "green",
      "internalLinks": "amber",
      "externalLinks": "green",
      "footnotes": "red",
      "references": "amber",
      "quotes": "red",
      "accuracy": "red"
    },
    "ratingsString": "N:5 R:7 A:5 C:7.5"
  },
  "readerRank": 496,
  "researchRank": 367,
  "recommendedScore": 215.75
}

{
  "lesswrong": "https://www.lesswrong.com/tag/eliciting-latent-knowledge",
  "stampy": "https://aisafety.info/questions/8Lfr/What-is-Eliciting-Latent-Knowledge-ELK",
  "alignmentForum": "https://www.alignmentforum.org/tag/eliciting-latent-knowledge"
}