{
  "docId": "019dd923-5e88-73ef-bd5c-f341d4394195",
  "docSlug": "46f66c49fd159048",
  "documentTitle": "2018 Air Street Capital The State of AI Report 2018",
  "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": 38,
  "pageCount": 156,
  "prevPage": 37,
  "nextPage": 39,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "dense",
  "nDataPoints": 12,
  "notes": "The slide includes a screenshot of the game interface and a comparison table of training infrastructure and performance metrics.",
  "elementsJson": [
    "headline_text",
    "paragraph",
    "screenshot",
    "comparison_table"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5c-f341d4394195/38",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5c-f341d4394195",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5c-f341d4394195.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5c-f341d4394195#slide-38",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "That RL agents can collaborate in teams to beat teams of humans is notable given the space of possible actions agents can take and the large maps they can interact with.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a80f-82a9120f0563",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.55,
        "x": 0.05,
        "y": 0.45
      },
      "kind": "image",
      "text": "Dota 2 gameplay screenshot with UI overlays",
      "attrs": null,
      "subkind": "screenshot",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "9471560a-7cfa-4142-af82-58b62125c34b",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.9,
        "x": 0.05,
        "y": 0.22
      },
      "kind": "paragraph",
      "text": "OpenAI Five is a team of 5 agents that learn through RL-based self-play to optimize their gameplay policy. The agents each have their own neural networks trained through RL to yield long-term planning behavior in a gameplay environment that is partially-observable and high-dimensional. That RL agents can collaborate in teams to beat teams of humans is notable given the space of possible actions agents can take and the large maps they can interact with.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "fdeca259-e2b6-45f7-b750-87209ab314dd",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.32,
        "x": 0.63,
        "y": 0.45
      },
      "kind": "table",
      "text": "Comparison table of OpenAI 1v1 Bot vs OpenAI Five infrastructure and training metrics",
      "attrs": null,
      "subkind": "data",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "a9278b55-815d-4484-b823-7dac3575dcb1",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.9,
        "x": 0.05,
        "y": 0.13
      },
      "kind": "title",
      "text": "OpenAI's multi-agent RL system learns to play complex real-time strategy game, Dota2",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "511b208a-29a4-48f8-9484-c2130ac619ea",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8eca-c39aa2ae7b9a",
      "evidence": "Title 'OpenAI's multi-agent RL system learns to play Dota2'.",
      "confidence": 85
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 69,
      "from": 4,
      "beatId": "019dd95a-0682-776c-8e34-ad4df4fe3ce7",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Definitions then research breakthroughs (transfer learning, hardware, RL) and talent supply data.",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 55,
      "from": 10,
      "beatId": "019dd95a-0682-776c-8e34-be65c7627fe7",
      "arcName": "The Onion",
      "arcSlug": "onion",
      "beatName": "First Layer",
      "beatSlug": "onion-first-layer",
      "evidence": "Hardware, vision, RL, bias - technical research layer.",
      "position": 2,
      "confidence": 55,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 39,
      "from": 36,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3c-55bec48734a5",
      "evidence": "AlphaZero, OpenAI Dota2, world-model RL — three parallel evidence pieces of RL progress.",
      "position": 5,
      "objective": "Show RL agents now beat human experts across domains",
      "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
}