{
  "docId": "019dd923-5e88-73ef-bd5d-01cf0d8a8fbc",
  "docSlug": "cc8183ec02431b7a",
  "documentTitle": "2022 Air Street Capital The State of AI Report 2022",
  "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": 103,
  "pageCount": 114,
  "prevPage": 102,
  "nextPage": 104,
  "slideType": "industry_trends",
  "function": "summarize",
  "density": "dense",
  "nDataPoints": 0,
  "notes": "The slide discusses the reverse-engineering of neural networks into human-interpretable programs.",
  "elementsJson": [
    "headline_text",
    "paragraph",
    "bullet_list",
    "infographic"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-01cf0d8a8fbc/103",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-01cf0d8a8fbc",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-01cf0d8a8fbc.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-01cf0d8a8fbc#slide-103",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Interpreting deep neural networks when seen just as sequences of matrix operations is exceedingly difficult – research in mechanistic interpretability instead seeks to reverse-engineer models into human-interpretable computer programs.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c4-7719-a21e-df4efa30adc1",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.35,
        "w": 0.35,
        "x": 0.63,
        "y": 0.48
      },
      "kind": "image",
      "text": "Induction heads implement the pattern [A][B]...[A]→[B] using prefix-matching and copying",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "902a6262-f521-4841-bfe7-31f34af7dd89",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.5,
        "w": 0.6,
        "x": 0.02,
        "y": 0.38
      },
      "kind": "list",
      "text": "Researchers at Anthropic released significant analyses of small transformer-based language models, focusing on a phenomenon of “induction heads” that learn to copy and complete sequences which have occurred before in a text. They find that these heads emerge during “phase shifts” in training during which in-context learning capabilities also emerge, and further developed a hypothesis that these heads may be responsible for the majority of in-context learning capabilities in large transformer models as well.\nFollow up work in this space has also brought to light ways in which individual neurons become responsible for individual or multiple semantic features, and ways to control this type of interpretability.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d31fec64-0b33-4d7e-963f-42543e6d3ef7",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.6,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "paragraph",
      "text": "Interpreting deep neural networks when seen just as sequences of matrix operations is exceedingly difficult – research in mechanistic interpretability instead seeks to reverse-engineer models into human-interpretable computer programs, gaining an understanding of individual neurons as well as their collective behaviour.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "708becda-9b8d-4203-847e-9e5c8f32a535",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.6,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Mechanistic interpretability – can we reverse-engineer neural networks?",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "94be1843-2bb6-4d6d-8ebb-40c893136b4d",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 106,
      "from": 81,
      "beatId": "019dd95a-0682-776c-8e35-28fc5bded874",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Implications (So What)",
      "beatSlug": "triple-take-the-implications-so-what",
      "evidence": "Politics and Safety sections interpret what the facts mean for governance and risk.",
      "position": 2,
      "confidence": 78,
      "parentBeatName": "Reflection",
      "parentBeatSlug": "reflection"
    },
    {
      "to": 106,
      "from": 94,
      "beatId": "019dd95a-0682-776c-8e35-381428a808af",
      "arcName": "The Mountain",
      "arcSlug": "mountain",
      "beatName": "Climax",
      "beatSlug": "mountain-climax",
      "evidence": "Safety section confronts existential risk and alignment urgency.",
      "position": 3,
      "confidence": 55,
      "parentBeatName": "Turn",
      "parentBeatSlug": "turn"
    }
  ],
  "loops": [
    {
      "to": 106,
      "from": 100,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3d-db1f31ffae02",
      "evidence": "RLHF, red-teaming, mechanistic interpretability, goal misgeneralization, moral behavior, Conjecture lab.",
      "position": 17,
      "objective": "Catalogue alignment-research techniques and example labs",
      "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
}