{
  "docId": "019dd923-5e88-73ef-bd5d-0d0f98caffe1",
  "docSlug": "5df6bd1b0447b5f6",
  "documentTitle": "2025 Air Street Capital The State of AI Report 2025",
  "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": 20,
  "pageCount": 313,
  "prevPage": 19,
  "nextPage": 21,
  "slideType": "diagnosis",
  "function": "diagnose",
  "density": "overcrowded",
  "nDataPoints": 12,
  "notes": "The chart shows 'Reported vs Measured Results' and 'Variance DS-R1-1.5B' to demonstrate that performance swings are often due to variance rather than actual model improvement.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "bar_chart_grouped",
    "box_plot"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0d0f98caffe1/20",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1#slide-20",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "The improvements we observe from recent reasoning methods fall entirely within baseline model variance ranges (i.e. margin of error), which suggests that perceived reasoning progress may be illusory.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a81b-3f1582be627c",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.35,
        "w": 0.42,
        "x": 0.53,
        "y": 0.28
      },
      "kind": "chart",
      "text": "Reported vs Measured Results and Variance DS-R1-1.5B",
      "attrs": null,
      "subkind": "bar-grouped",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "48b29bde-b991-43d0-8269-c5f0b72a1aea",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.45,
        "x": 0.05,
        "y": 0.51
      },
      "kind": "list",
      "text": "What's more, RL approaches show minimal real gains and overfit easily. Under standardised evaluation, many RL methods drop 6-17% from reported results with no statistically significant improvements over baselines.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "43aaa6f2-525f-4c73-ab8c-684412cef186",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.18,
        "w": 0.45,
        "x": 0.05,
        "y": 0.31
      },
      "kind": "list",
      "text": "Current benchmarks are highly sensitive to the implementation (decoding parameters, seeds, prompts, hardware) and small dataset sizes. For example, AIME'24 only has 30 examples where one question shifts Pass@1 by 3+ percentage points, causing double-digit performance swings.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "a48c9bb1-1c5c-44b6-b00b-979befbc073a",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.45,
        "x": 0.05,
        "y": 0.65
      },
      "kind": "list",
      "text": "Recent methods' improvements fall entirely within baseline model variance ranges, suggesting limited progress. This highlights the critical need for rigorous multi-seed evaluation protocols and transparent reporting standards.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f8f445cc-f15f-4e67-95f2-d8272f1ca27a",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Accuracy (%)",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a81b-405a65e631c9",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.45,
        "x": 0.05,
        "y": 0.21
      },
      "kind": "paragraph",
      "text": "The improvements we observe from recent reasoning methods fall entirely within baseline model variance ranges (i.e. margin of error), which suggests that perceived reasoning progress may be illusory...",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d4f4cb6c-bc97-4536-b047-112db0d553f3",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.35,
        "x": 0.05,
        "y": 0.14
      },
      "kind": "title",
      "text": "The illusion of reasoning gains",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "b8c8333c-7282-4120-970d-605ab326ea39",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a95b-ec992c79dfb4",
      "evidence": "'The illusion of reasoning gains' delivers the headline insight.",
      "confidence": 90
    },
    {
      "name": "Unexpected Pattern",
      "slug": "unexpected-pattern",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a95b-f1dd5dcaf612",
      "evidence": "Counter-intuitive claim that reasoning gains fall within margin of error.",
      "confidence": 80
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 189,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-9f380673831f",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Sections 1-2 lay out research findings and industry data with charts and case studies.",
      "position": 1,
      "confidence": 70,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 89,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-ac68557d958b",
      "arcName": "The Onion",
      "arcSlug": "onion",
      "beatName": "First Layer",
      "beatSlug": "onion-first-layer",
      "evidence": "Research section unpacks technical layer beneath the headlines.",
      "position": 2,
      "confidence": 45,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 30,
      "from": 13,
      "name": "Zoom In",
      "slug": "06-zoom-in",
      "bestFor": "Technical deep-dives, case studies, detailed analysis",
      "matchId": "019dd95a-07fe-70ce-8d3e-9252f6f92bd7",
      "evidence": "Big picture 'reasoning race' (p.13) zooms into o1, R1, DeepSeek, parallel reasoning, CoT mechanics.",
      "position": 2,
      "objective": "Drill from reasoning paradigm into specific reasoning research findings",
      "structure": "The Big Picture -> Key Area of Focus -> Specific Detail -> Implication",
      "confidence": 75,
      "description": "Start broad, then progressively focus on specific details that prove your point"
    },
    {
      "to": 23,
      "from": 20,
      "name": "Myth Buster",
      "slug": "12-myth-buster",
      "bestFor": "Rebranding, changing market perception, correcting false assumptions",
      "matchId": "019dd95a-07fe-70ce-8d3e-94ced6245fc4",
      "evidence": "'The illusion of reasoning gains' (p.20), 'How reasoning breaks' (pp.22-23) directly challenge consensus.",
      "position": 3,
      "objective": "Bust the 'reasoning gains are real' belief",
      "structure": "The Common Belief -> The Friction/Failure of that Belief -> The New Truth",
      "confidence": 80,
      "description": "Address a common misconception head-on to clear the room for a new truth"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}