{
  "docId": "019dd923-5e88-73ef-bd5c-fd5012384ae3",
  "docSlug": "bf350dd574c19997",
  "documentTitle": "2021 Air Street Capital The State of AI Report 2021",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 44,
  "pageCount": 188,
  "prevPage": 43,
  "nextPage": 45,
  "slideType": "diagnosis",
  "function": "diagnose",
  "density": "overcrowded",
  "nDataPoints": 24,
  "notes": "Includes a qualitative example of model responses to tricky questions as model size increases.",
  "elementsJson": [
    "headline_text",
    "action_title",
    "bullet_list",
    "bar_chart_vertical",
    "comparison_table"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5c-fd5012384ae3/44",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5c-fd5012384ae3#slide-44",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "More surprisingly, models of larger sizes were generally less truthful.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d27-a3d42a670695",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.2,
        "w": 0.45,
        "x": 0.02,
        "y": 0.5
      },
      "kind": "chart",
      "text": "Figure 1 : Average truthfulness on Truthful QA",
      "attrs": null,
      "subkind": "bar-vertical",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "20483c06-c2dc-4d29-8117-3ab9eeba8f00",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.2,
        "w": 0.45,
        "x": 0.02,
        "y": 0.75
      },
      "kind": "chart",
      "text": "Figure 2: Average truthfulness on control trivia questions",
      "attrs": null,
      "subkind": "bar-vertical",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "adab0bb2-33ca-4848-891b-d2cd19da8e6e",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.95,
        "x": 0.02,
        "y": 0.38
      },
      "kind": "list",
      "text": "While LLMs were relatively truthful on control trivia questions, they struggled on TruthfulQA, which contains questions which were designed to fool the largest GPT-3.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "7e21cc84-33b6-4c9e-aa5d-c9f0977e6a2c",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Truthfulness percentage: 58%",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d27-a5eda1fef12d",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.95,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "paragraph",
      "text": "Researchers tested large language models on TruthfulQA, a new benchmark of questions spanning domains such as health, law, conspiracies and fiction. They showed that the best model was truthful on 58% of the questions, compared to the human baseline of 94%. More surprisingly, models of larger sizes were generally less truthful.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "3819ef75-41be-42de-876e-20c9718d1647",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.35,
        "w": 0.45,
        "x": 0.5,
        "y": 0.5
      },
      "kind": "table",
      "text": "Qualitative comparison of model responses by size",
      "attrs": null,
      "subkind": "data",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "ff35f327-2e91-4c60-9c68-cf80a6dd3331",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.7,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Big fat liars: large language models are less truthful than their smaller peers",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f63c25e4-c7e8-42ca-925d-e26b9117d854",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Opening Hooks",
      "slug": "opening-hooks",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8ed1-10e0986ad2d2",
      "evidence": "Hook 'Big fat liars' is a provocative startle.",
      "confidence": 78
    },
    {
      "name": "Unexpected Pattern",
      "slug": "unexpected-pattern",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8ed1-153d2172fb99",
      "evidence": "Larger models LESS truthful — counterintuitive finding.",
      "confidence": 80
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 152,
      "from": 5,
      "beatId": "019dd95a-0682-776c-8e35-0affbadec38e",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Definitions p.5-6, Exec Summary p.7, then Sections 1-3 catalog evidence.",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 152,
      "from": 10,
      "beatId": "019dd95a-0682-776c-8e35-1b3ee3a1c012",
      "arcName": "The Mountain",
      "arcSlug": "mountain",
      "beatName": "Rising Action",
      "beatSlug": "mountain-rising-action",
      "evidence": "Three sections accumulating evidence of accelerating progress.",
      "position": 2,
      "confidence": 55,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 51,
      "from": 41,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3d-576c8c35c986",
      "evidence": "Codex, math limits, truthfulness, prompt engineering, SuperGLUE, multilingual LLMs.",
      "position": 6,
      "objective": "Trace LLM frontier and the prompting paradigm",
      "structure": "Evidence A -> Evidence B -> Evidence C -> Pattern/Conclusion",
      "confidence": 78,
      "description": "Group multiple pieces of evidence that together point to a pattern or conclusion"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}