{
  "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": 67,
  "pageCount": 213,
  "prevPage": 66,
  "nextPage": 68,
  "slideType": "industry_trends",
  "function": "summarize",
  "density": "dense",
  "nDataPoints": 3,
  "notes": "The slide features a process diagram showing a booking flow with failure and success states.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "process_diagram",
    "screenshot"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0856e1444fb9/67",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9#slide-67",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "They found this improved Llama-3 70B’s zero-shot performance from 18.6% to 81.7% in real-world booking scenarios, after a day of data collection, and up to 95.4% with online search.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d14-778a3de8348e",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.4,
        "w": 0.5,
        "x": 0.45,
        "y": 0.5
      },
      "kind": "image",
      "text": "Process diagram showing a booking flow with failure and success states.",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "0dd07c39-cb8f-4e74-9000-8e1fe987e978",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.6,
        "w": 0.45,
        "x": 0.05,
        "y": 0.3
      },
      "kind": "list",
      "text": "MultiOn and Stanford combined an LLM with MCTS, along with a self-criticism mechanism and direct preference optimization, to learn from different success and failure criteria.\nThey found this improved Llama-3 70B's zero-shot performance from 18.6% to 81.7% in real-world booking scenarios, after a day of data collection, and up to 95.4% with online search.\nThe longer-term question will be whether next-token prediction loss is too fine-grained.\nThis risks limiting the ability of RL and MCTS to achieve agentic behavior by focusing too much on individual tokens and hindering the exploration of broader, more strategic solutions.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "65c824db-e925-45af-bedc-45ff9ca2a995",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "zero-shot performance: 95.4%",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d14-7a5960db5216",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.9,
        "x": 0.05,
        "y": 0.2
      },
      "kind": "paragraph",
      "text": "To improve planning, approaches like MCTS, which helped to power AlphaGo, are slowly returning to the fore. Early results are promising, but will they be enough?",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "74477251-be8d-44f3-b49c-1e43fe951951",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.4,
        "x": 0.05,
        "y": 0.13
      },
      "kind": "title",
      "text": "Who remembers Monte Carlo Tree Search?",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f7af6af4-ee15-418b-9ac8-9ef0b6ce1bde",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a959-28facaf89a2b",
      "evidence": "Title 'Who remembers Monte Carlo Tree Search?' uses rhetorical question",
      "confidence": 80
    }
  ],
  "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
}