{
  "docId": "019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "docSlug": "8757f1b44ef7f176",
  "documentTitle": "2024 Air Street Capital The State of AI Report 2024",
  "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": 68,
  "pageCount": 213,
  "prevPage": 67,
  "nextPage": 69,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "dense",
  "nDataPoints": 1,
  "notes": "The slide highlights the transition from research to application (indicated by the pencil-to-wrench icon).",
  "elementsJson": [
    "bullet_list",
    "paragraph",
    "screenshot"
  ],
  "metadataConfidence": 0.95,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0856e1444fb9/68",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9#slide-68",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Genie is both able to imagine entirely new interactive scenes and demonstrate significant flexibility: it can take prompts in various forms, from text descriptions to hand-drawn sketches, and bring them to life as playable environments.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d14-efe15effc424",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.4,
        "w": 0.3,
        "x": 0.65,
        "y": 0.45
      },
      "kind": "image",
      "text": "Visual demonstration of Genie: Text-to-Image, Hand-drawn Sketch, and Real-world Photo prompts resulting in playable environments.",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "30a014be-56bf-440f-8e7e-dedf4261247f",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.7,
        "w": 0.6,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "list",
      "text": "One of the big bottlenecks for training RL agents is a shortage of training data. Standard approaches like converting pre-existing environments (e.g. Atari) or manually building them are labor-intensive and don't scale.\nGenie (winner of a Best Paper award at ICML 2024) is a world model that can generate action-controllable virtual worlds. It analyzed 30,000 hours of video game footage from 2D platformer games, learning to compress the visual information and infer the actions that drive changes between frames.\nBy learning a latent action space from video data, it can handle action representations without requiring explicit action labels, which distinguishes it from other world models.\nGenie is both able to imagine entirely new interactive scenes and demonstrate significant flexibility: it can take prompts in various forms, from text descriptions to hand-drawn sketches, and bring them to life as playable environments.\nThis approach demonstrated applicability beyond games, with the team successfully applying the hyperparameters from the game model to robotics data, without fine tuning.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "92a3623a-1ec0-4574-a140-230c1bc74c69",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.6,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Could foundation models make it easier to train RL agents at scale?",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "8ef6fc91-fc2d-4f9a-9e07-804825c6ee1b",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 153,
      "from": 9,
      "beatId": "019dd95a-0682-776c-8e35-5f2398b8d1d0",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Exec summary + Research + Industry sections inventory the year",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 85,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-6d9d8ee06d0a",
      "arcName": "Voyage and Return",
      "arcSlug": "voyage-return",
      "beatName": "The Unknown",
      "beatSlug": "voyage-return-the-unknown",
      "evidence": "Research section explores frontier model uncharted territory",
      "position": 2,
      "confidence": 55,
      "parentBeatName": "Complication",
      "parentBeatSlug": "complication"
    }
  ],
  "loops": [
    {
      "to": 71,
      "from": 59,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3e-3f24b9b4c1e3",
      "evidence": "ARC challenge, planning failures, COT, open-endedness, MCTS, FunSearch, RL-LLM combos",
      "position": 8,
      "objective": "Catalogue progress and limits in reasoning research",
      "structure": "Evidence A -> Evidence B -> Evidence C -> Pattern/Conclusion",
      "confidence": 76,
      "description": "Group multiple pieces of evidence that together point to a pattern or conclusion"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}