{
  "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": 21,
  "pageCount": 213,
  "prevPage": 20,
  "nextPage": 22,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "dense",
  "nDataPoints": 2,
  "notes": "The slide highlights the hybrid approach of neuro-symbolic AI in solving Olympiad-level math problems.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "paragraph",
    "process_diagram",
    "screenshot"
  ],
  "metadataConfidence": 0.95,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0856e1444fb9/21",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9#slide-21",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Impressively, It solved 25 out of 30 on a benchmark of Olympiad-level geometry problems, nearing human International Mathematical Olympiad gold medalist performance.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d11-7f8aaad7fcbe",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.4,
        "x": 0.55,
        "y": 0.4
      },
      "kind": "image",
      "text": "AlphaGeometry process diagram showing IMO 2015 P3 problem and solution steps.",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f3df87a0-2844-497a-99c9-2c1e5c7ea97e",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.45,
        "x": 0.05,
        "y": 0.32
      },
      "kind": "list",
      "text": "A Google DeepMind/NYU team generated millions of synthetic theorems and proofs using symbolic engines, using them to train a language model from scratch.\nAlphaGeometry alternates between the language model proposing new constructions and symbolic engines performing deductions until a solution is found.\nImpressively, It solved 25 out of 30 on a benchmark of Olympiad-level geometry problems, nearing human International Mathematical Olympiad gold medalist performance. The next best AI performance scored only 10.\nIt also demonstrated generalisation capabilities - for example, finding that a specific detail in a 2004 IMO problem was unnecessary for the proof.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "41ca30c1-3876-487e-a3f8-3a882c8af0da",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Olympiad-level geometry problems solved: 25",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d11-8118869d7321",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.1,
        "w": 0.45,
        "x": 0.05,
        "y": 0.21
      },
      "kind": "paragraph",
      "text": "Deficiencies in both reasoning capabilities and training data mean that AI systems have frequently fallen short on math and geometry problems. With AlphaGeometry, a symbolic deduction engine comes to the rescue.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "371b7695-5f67-4349-8633-a4035c0767c7",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.45,
        "x": 0.05,
        "y": 0.13
      },
      "kind": "title",
      "text": "Are neuro-symbolic systems making a comeback?",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "2a7267a9-778a-492f-b233-2446507d91de",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Curiosity Gap",
      "slug": "curiosity-gap",
      "agent": "Storyteller",
      "layer": "block",
      "matchId": "019dd95a-1055-74e0-a958-895977bd4d8b",
      "evidence": "Question title creates a gap",
      "confidence": 70
    },
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a958-858bd539b43f",
      "evidence": "Title 'Are neuro-symbolic systems making a comeback?'",
      "confidence": 75
    },
    {
      "name": "Aha! Moment",
      "slug": "aha-moment",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "5912ecb1-933a-462e-8f5c-9774fb345022",
      "evidence": "Impressively, It solved 25 out of 30 on a benchmark of Olympiad-level geometry problems, nearing human International Mathematical Olympiad gold medalist performance.",
      "confidence": 0.9
    }
  ],
  "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": 29,
      "from": 21,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3e-2b2229583eee",
      "evidence": "Eight slides on shrinking models, on-device, hybrids, transformers reigning",
      "position": 3,
      "objective": "Demonstrate efficiency frontier across distillation, quantization, hybrids",
      "structure": "Evidence A -> Evidence B -> Evidence C -> Pattern/Conclusion",
      "confidence": 75,
      "description": "Group multiple pieces of evidence that together point to a pattern or conclusion"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}