{
  "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": 56,
  "pageCount": 213,
  "prevPage": 55,
  "nextPage": 57,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "dense",
  "nDataPoints": 3,
  "notes": "The slide details the architecture and performance benefits of the Aurora model compared to traditional numerical weather prediction.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "process_diagram",
    "paragraph"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0856e1444fb9/56",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9#slide-56",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "The model is also 5,000x faster than the Integrated Forecasting System that uses numerical forecasting.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d12-ee2484b4d31f",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.41,
        "x": 0.56,
        "y": 0.42
      },
      "kind": "image",
      "text": "Diagram showing the Aurora model architecture: Pretraining, Finetuning & Inference, 3D Perceiver Encoder, 3D Swin Transformer, and 3D Perceiver Decoder.",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "e4a837df-1701-4852-bfdc-8b1e2b635f87",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.52,
        "x": 0.03,
        "y": 0.42
      },
      "kind": "list",
      "text": "The 1.3B model is pre-trained on >1M hours of weather and climate data from 6 datasets, including forecasts, analysis data, reanalysis data, and climate simulations.\nThe models encodes heterogeneous inputs into a standard 3D representation of the atmosphere across space and pressure-levels, which is evolved over time at inference by a vision transformer and decoded into specific predictions.\nImportantly, it is the first model to predict atmospheric chemistry (6 major air pollutants, e.g. ozone, carbon monoxide), which involves hundreds of stiff equations, better than numerical models. The model is also 5,000x faster than the Integrated Forecasting System that uses numerical forecasting.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "b3508776-106a-4684-b9d7-0b09b83c8251",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Computational speed: 5,000x",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d12-f0c4ad5b11d5",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.18,
        "w": 0.94,
        "x": 0.03,
        "y": 0.21
      },
      "kind": "paragraph",
      "text": "Classical atmospheric simulation methods like numerical weather prediction are costly and unable to make use of diverse and often scarce atmospheric data modalities. But, foundation models are well suited here. Microsoft researchers created Aurora, a foundation model that produces forecasts for a wide range of atmospheric forecasting problems such as global air pollution and high-resolution medium-term weather patterns. It can also adapt to new tasks by making use of a general-purpose learned representation of atmospheric dynamics.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "eb76bc93-9f14-433f-8a4f-e60f45972bc7",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.55,
        "x": 0.03,
        "y": 0.13
      },
      "kind": "title",
      "text": "Foundation models across the sciences: the atmosphere",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d881f0ee-8ae9-41a8-8c9b-663c2f43b224",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a958-f8dc8bf92086",
      "evidence": "Title 'Foundation models across the sciences: the atmosphere'",
      "confidence": 75
    },
    {
      "name": "Weber-Fechner Law",
      "slug": "weber-fechner-law",
      "agent": "Designer",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a958-fdc9d698d199",
      "evidence": "Callout '5,000x faster' as huge magnitude leverages numeric contrast",
      "confidence": 50
    }
  ],
  "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": 58,
      "from": 45,
      "name": "Zoom In",
      "slug": "06-zoom-in",
      "bestFor": "Technical deep-dives, case studies, detailed analysis",
      "matchId": "019dd95a-07fe-70ce-8d3e-3bc439910222",
      "evidence": "Nobels, AlphaFold3, AlphaProteo, ESM3, gene editors, materials, brain fMRI, speech-from-cortex case studies",
      "position": 7,
      "objective": "Zoom into AI for science domain by domain",
      "structure": "The Big Picture -> Key Area of Focus -> Specific Detail -> Implication",
      "confidence": 80,
      "description": "Start broad, then progressively focus on specific details that prove your point"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}