{
  "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": 52,
  "pageCount": 213,
  "prevPage": 51,
  "nextPage": 53,
  "slideType": "problem_statement",
  "function": "frame_problem",
  "density": "dense",
  "nDataPoints": 24,
  "notes": "Includes logos of PoseCheck, PoseBusters, Polaris, and Inductive Bio at the bottom.",
  "elementsJson": [
    "headline_text",
    "action_title",
    "bullet_list",
    "bar_chart_grouped",
    "paragraph"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0856e1444fb9/52",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0856e1444fb9#slide-52",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "The fundamental problem with research at the intersection of biology and ML is that there are very few people with the skills to both train a frontier model and give it a rigorous biological appraisal.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d13-28def0d1acbe",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.4,
        "w": 0.38,
        "x": 0.6,
        "y": 0.45
      },
      "kind": "chart",
      "text": "Performance comparison of various docking methods across Astex and PoseBusters benchmarks.",
      "attrs": null,
      "subkind": "bar-grouped",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "86cca6e3-c951-494f-87c2-078205c82234",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.65,
        "w": 0.55,
        "x": 0.03,
        "y": 0.22
      },
      "kind": "list",
      "text": "The fundamental problem with research at the intersection of biology and ML is that there are very few people with the skills to both train a frontier model and give it a rigorous biological appraisal.\nTwo works from late 2023, PoseCheck and PoseBusters, showed that ML models for molecule generation and protein-ligand docking gave structures (poses) with gross physical violations.\nEven the AlphaFold3 paper didn't get away without a few bruises when Inductive bio showed that using a slightly more advanced conventional docking pipeline beat AF3.\nA new industry consortium led by Valence Labs, including major pharma companies (i.e. Recursion, Relay, Merck, Novartis J&J, Pfizer), is developing Polaris, a benchmarking platform for AI-driven drug discovery. Polaris will provide high-quality datasets, facilitate evaluations, and certify benchmarks.\nMeanwhile, Recursion's work on perturbative map-building led them to create a new set of benchmarks and metrics.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "98136afa-a645-42cf-8ddd-c2bffbc21eff",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Percentage of predictions",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c3-7407-8d13-2d313188517a",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.45,
        "x": 0.03,
        "y": 0.14
      },
      "kind": "title",
      "text": "Yet, evals and benchmarking in BioML remains poor",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "0e87267c-5323-41a9-ad92-4e7e767ecdfe",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-1055-74e0-a958-f6ea00193f11",
      "evidence": "Title 'Yet, evals and benchmarking in BioML remains poor'",
      "confidence": 80
    },
    {
      "name": "Problem Statement Canvas",
      "slug": "problem-statement-canvas",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "d09d71eb-4138-483a-9f0b-8970f2039863",
      "evidence": "The fundamental problem with research at the intersection of biology and ML is that there are very few people with the skills to both train a frontier model and give it a rigorous biological appraisal",
      "confidence": 0.6
    }
  ],
  "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
}