{
  "version": "2.0",
  "publisher": "AI Nextia",
  "canonical": "https://ainextia.com/advancements/",
  "reviewed": "2026-11-07",
  "recordCount": 111,
  "description": "Curated source-attributed milestones, observed impacts, studies and releases. Not independently replicated by AI Nextia.",
  "methodology": [
    "Each entry describes a distinct capability, reported outcome, study finding or historical release; none is a claim that AI Nextia independently reproduced the result.",
    "A product release means a tool exists or was demonstrated, not that it is universally available or a proven productivity improvement.",
    "Operational deployments and independently measured studies are labeled separately from laboratory demonstrations and vendor claims.",
    "Year-wide 2025 statistics are grouped under 2025 Annual Findings without fabricating an exact announcement month.",
    "Dates of individual releases are taken from original sources where possible; historical records are not proof of current product availability.",
    "Source-linked adoption figures and modeled forecasts retain their original geographic, sample and methodological limitations.",
    "The 2027+ watchlist is not a list of achievements or promised release dates.",
    "The catalog is curated and growing; it is not an exhaustive index of every AI model, paper, company or social-media demonstration."
  ],
  "records": [
    {
      "id": "adv-024",
      "date": "2026-11-05",
      "precision": "day",
      "year": 2026,
      "month": "2026-11",
      "title": "Specialized AI assists authorized cybersecurity work",
      "category": "Safety & Security",
      "kind": "Restricted release",
      "access": "Restricted access",
      "organization": "OpenAI",
      "finding": "GPT-6-Cyber was released for authorized vulnerability research and remediation verification.",
      "possible": "Authorized security teams can test controlled defensive investigation and fix verification.",
      "caution": "Restricted access and authorized environments are required; not consumer-facing unrestricted use.",
      "source": "https://developers.openai.com/api/docs/changelog",
      "keywords": "cybersecurity bug vulnerability authorized",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-023",
      "date": "2026-11-04",
      "precision": "day",
      "year": 2026,
      "month": "2026-11",
      "title": "AI participates in a multi-step biology discovery workflow",
      "category": "Science",
      "kind": "Company case study",
      "access": "Research-stage",
      "organization": "Anthropic",
      "finding": "Anthropic documented an AI-assisted workflow for studying vault-like particles using biological tools and cryo-EM.",
      "possible": "Explore research assistance across literature, protocol drafting and image analysis with scientists in control.",
      "caution": "A single case study does not establish universal scientific accuracy.",
      "source": "https://www.anthropic.com/research/investigating-vault-particles-with-claude",
      "keywords": "biology cryoem science research",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-022",
      "date": "2026-11-04",
      "precision": "day",
      "year": 2026,
      "month": "2026-11",
      "title": "Animated GIF generation arrives in image workflows",
      "category": "Creative AI",
      "kind": "Product release",
      "access": "Available with limits",
      "organization": "OpenAI",
      "finding": "Images 2.6 added animated GIF creation, stronger iterative editing and image controls.",
      "possible": "Create animated concepts, brand mascots and lightweight demonstrations with review.",
      "caution": "Generated animations may contain temporal or factual errors; usage rights still matter.",
      "source": "https://openai.com/index/introducing-images-2-6/",
      "keywords": "images animation gif creative",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-020",
      "date": "2026-11-03",
      "precision": "day",
      "year": 2026,
      "month": "2026-11",
      "title": "AI-agent chemistry demonstration spans compound search and design",
      "category": "Biology & Medicine",
      "kind": "Research demonstration",
      "impactMetric": "One-hour demonstrated cheminformatics workflow",
      "finding": "DeepMind showed an agent comparing compounds and proposing molecule structures and synthesis routes.",
      "possible": "Drug researchers can streamline literature and chemical-data exploration.",
      "caution": "Proposed candidates are not synthesized or proven medicines.",
      "source": "https://deepmind.google/blog/agentic-cheminformatics-with-google-antigravity/",
      "organization": "Google DeepMind",
      "access": "Research demonstration",
      "keywords": "ai-agent chemistry demonstration spans compound search and design biology & medicine research demonstration one-hour demonstrated cheminformatics workflow",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-021",
      "date": "2026-11-02",
      "precision": "day",
      "year": 2026,
      "month": "2026-11",
      "title": "Research explores cryptographic proof of model inference",
      "category": "Infrastructure",
      "kind": "Research prototype",
      "access": "Experimental",
      "organization": "Anthropic",
      "finding": "Research on provable inference explores proving which model served a result without revealing model weights.",
      "possible": "Follow potential provenance controls for enterprise AI deployments.",
      "caution": "Not an ordinary production verification feature; overhead and deployment remain open questions.",
      "source": "https://www.anthropic.com/research/provable-inference",
      "keywords": "verification cryptography AI model",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-025",
      "date": "2026-11",
      "precision": "month",
      "year": 2026,
      "month": "2026-11",
      "title": "Agentic cheminformatics explores scientific tooling",
      "category": "Science",
      "kind": "Research announcement",
      "access": "Research-stage",
      "organization": "Google DeepMind",
      "finding": "DeepMind listed November research on agentic chemistry information processing.",
      "possible": "Watch how scientific agents connect literature, molecular data and validated tools.",
      "caution": "Headline-level evidence only; research claims require paper-level review.",
      "source": "https://deepmind.google/blog/",
      "keywords": "chemistry scientific agents",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-026",
      "date": "2026-11",
      "precision": "month",
      "year": 2026,
      "month": "2026-11",
      "title": "Gemini Omni 1.2 Flash expands multimedia model options",
      "category": "Models",
      "kind": "Product announcement",
      "access": "Product-dependent",
      "organization": "Google DeepMind",
      "finding": "DeepMind listed a November update focused on faster multimedia generation and longer clips.",
      "possible": "Compare supported multimedia tasks and costs on your actual workflow.",
      "caution": "Month-level announcement; exact supported features require current docs.",
      "source": "https://deepmind.google/blog/",
      "keywords": "video audio multimodal clips",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-020",
      "date": "2026-10-26",
      "precision": "day",
      "year": 2026,
      "month": "2026-10",
      "title": "GPT-6.1 Terra extends computer use at scale",
      "category": "Agents",
      "kind": "Model release",
      "access": "Product-dependent",
      "organization": "OpenAI",
      "finding": "OpenAI's release archive describes Terra as a model focused on computer-use tasks.",
      "possible": "Benchmark real browser and desktop actions with explicit permission gates.",
      "caution": "Model availability and task reliability depend on access and environment.",
      "source": "https://openai.com/research/index/release/",
      "keywords": "computer use desktop agent",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-021",
      "date": "2026-10-15",
      "precision": "day",
      "year": 2026,
      "month": "2026-10",
      "title": "AI research agents prioritize antimicrobial peptides",
      "category": "Biology & Medicine",
      "kind": "Research demonstration",
      "impactMetric": "Lab validation pending",
      "finding": "DeepMind described an agent identifying a plausible antimicrobial peptide through protein data searches and structure analysis.",
      "possible": "Researchers can narrow a huge search space before expensive experiments.",
      "caution": "Predicted biological activity is not experimentally demonstrated efficacy.",
      "source": "https://deepmind.google/blog/accelerating-antimicrobial-discovery/",
      "organization": "Google DeepMind",
      "access": "Research demonstration",
      "keywords": "ai research agents prioritize antimicrobial peptides biology & medicine research demonstration lab validation pending",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "impact-013",
      "date": "2026-10-05",
      "precision": "day",
      "year": 2026,
      "month": "2026-10",
      "title": "PIE predicts cell responses in unseen conditions",
      "category": "Biology & Medicine",
      "kind": "Research release",
      "impactMetric": "Out-of-context perturbation generalization",
      "finding": "Arc's PIE model uses biological priors to improve prediction of population-level cell responses.",
      "possible": "Scientists can prioritize experiments in previously unseen settings.",
      "caution": "Predictions need withheld-data testing and experimental replication.",
      "source": "https://arcinstitute.org/news/pie",
      "organization": "Arc Institute",
      "access": "Research release",
      "keywords": "pie predicts cell responses in unseen conditions biology & medicine research release out-of-context perturbation generalization",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-019",
      "date": "2026-10",
      "precision": "month",
      "year": 2026,
      "month": "2026-10",
      "title": "Genie 4 advances interactive world simulation",
      "category": "Creative AI",
      "kind": "Research announcement",
      "access": "Research-stage",
      "organization": "Google DeepMind",
      "finding": "Google DeepMind announced a new generation of world-simulation research.",
      "possible": "Watch for simulation-based prototyping and interactive scene exploration.",
      "caution": "World simulations are not exact models of the physical world.",
      "source": "https://deepmind.google/blog/",
      "keywords": "simulation worlds video",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-016",
      "date": "2026-09-08",
      "precision": "day",
      "year": 2026,
      "month": "2026-09",
      "title": "AlphaGenome Atlas maps 9 billion possible DNA variants",
      "category": "Science",
      "kind": "Research dataset",
      "access": "Research portal",
      "organization": "Google DeepMind",
      "finding": "The Atlas published predicted molecular effects for roughly nine billion possible single-letter variants.",
      "possible": "Explore questions about gene regulation and candidate variants in research settings.",
      "caution": "Predictions are computational, not validated clinical conclusions.",
      "source": "https://deepmind.google/blog/alphagenome-atlas-a-predictive-map-of-every-possible-dna-letter-change-in-the-human-genome/",
      "keywords": "dna genome atlas biology research",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-023",
      "date": "2026-09-08",
      "precision": "day",
      "year": 2026,
      "month": "2026-09",
      "title": "Nine billion DNA changes receive predictive annotations",
      "category": "Biology & Medicine",
      "kind": "Research platform",
      "impactMetric": "Nine billion possible single-letter variants",
      "finding": "DeepMind released AlphaGenome Atlas predictions at human-genome scale.",
      "possible": "Researchers can search a broad set of candidate molecular effects.",
      "caution": "These are modeled changes rather than billions of laboratory observations.",
      "source": "https://deepmind.google/blog/alphagenome-atlas-a-predictive-map-of-every-possible-dna-letter-change-in-the-human-genome/",
      "organization": "Google DeepMind",
      "access": "Research platform",
      "keywords": "nine billion dna changes receive predictive annotations biology & medicine research platform nine billion possible single-letter variants",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-018",
      "date": "2026-09",
      "precision": "month",
      "year": 2026,
      "month": "2026-09",
      "title": "Agentic video understanding enters model workflows",
      "category": "Creative AI",
      "kind": "Product announcement",
      "access": "Product-dependent",
      "organization": "Google DeepMind",
      "finding": "DeepMind announced agentic video understanding capabilities among its September model updates.",
      "possible": "Explore video search, scene analysis and clip retrieval subject to permissions.",
      "caution": "Feature availability and accuracy depend on provider and implementation.",
      "source": "https://deepmind.google/blog/",
      "keywords": "video analysis multimodal",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-009",
      "date": "2026-09",
      "precision": "month",
      "year": 2026,
      "month": "2026-09",
      "title": "Robotaxis accumulate hundreds of millions of driverless miles",
      "category": "Mobility & Transport",
      "kind": "Operational evidence",
      "impactMetric": "271.3 million rider-only miles through June 2026",
      "finding": "Waymo published city-level mileage and safety-comparison methodology.",
      "possible": "Large driving datasets support empirical safety evaluation rather than relying on demo clips.",
      "caution": "Reported safety comparisons depend on exposure, location and benchmark definitions.",
      "source": "https://waymo.com/safety/impact/",
      "organization": "Waymo",
      "access": "Operational evidence",
      "keywords": "robotaxis accumulate hundreds of millions of driverless miles mobility & transport operational evidence 271.3 million rider-only miles through june 2026",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "adv-017",
      "date": "2026-09",
      "precision": "month",
      "year": 2026,
      "month": "2026-09",
      "title": "WeatherNext 3 expands AI weather-model research",
      "category": "Science",
      "kind": "Research announcement",
      "access": "Research-stage",
      "organization": "Google DeepMind",
      "finding": "DeepMind announced a new generation of WeatherNext global weather forecasting models.",
      "possible": "Investigate research-led weather modeling and uncertainty-aware forecast workflows.",
      "caution": "Forecast performance depends on region, horizon and evaluation; never replace official warnings.",
      "source": "https://deepmind.google/blog/",
      "keywords": "weather forecast science",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-006",
      "date": "2026-08-18",
      "precision": "day",
      "year": 2026,
      "month": "2026-08",
      "title": "Flash-flood prediction expands from public disaster records",
      "category": "Climate & Weather",
      "kind": "Research deployment",
      "impactMetric": "150-country flood forecasting coverage",
      "finding": "Google documented wider flood forecasting and Groundsource methodology for extracting flash-flood events.",
      "possible": "Researchers can build better geographically varied flood-risk baselines.",
      "caution": "Model predictions are not official evacuation instructions.",
      "source": "https://blog.google/innovation-and-ai/technology/research/flood-prediction-ai/",
      "organization": "Google",
      "access": "Research deployment",
      "keywords": "flash-flood prediction expands from public disaster records climate & weather research deployment 150-country flood forecasting coverage",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-015",
      "date": "2026-07-09",
      "precision": "day",
      "year": 2026,
      "month": "2026-07",
      "title": "ChatGPT Work supports longer multi-app tasks",
      "category": "Agents",
      "kind": "Product release",
      "access": "Plan-dependent",
      "organization": "OpenAI",
      "finding": "ChatGPT Work introduced agents that operate across web, files and apps to complete more substantial projects.",
      "possible": "Test multi-source research, document production and cross-tool business processes.",
      "caution": "Control permissions, verify outputs and check rollout availability.",
      "source": "https://openai.com/index/chatgpt-for-your-most-ambitious-work/",
      "keywords": "agent office apps business workflows",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-037",
      "date": "2026-06-16",
      "precision": "day",
      "year": 2026,
      "month": "2026-06",
      "title": "Coding-agent use expands beyond software repairs",
      "category": "Software & Work",
      "kind": "Observational study",
      "impactMetric": "Around 400,000 sessions analyzed",
      "finding": "Anthropic documented coding agents increasingly supporting operations, writing and analysis tasks.",
      "possible": "Organizations can evaluate expanding use cases across teams.",
      "caution": "Usage growth is not controlled proof of productivity.",
      "source": "https://www.anthropic.com/research/claude-code-expertise",
      "organization": "Anthropic",
      "access": "Observational study",
      "keywords": "coding-agent use expands beyond software repairs software & work observational study around 400,000 sessions analyzed",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend"
    },
    {
      "id": "adv-014",
      "date": "2026-06-16",
      "precision": "day",
      "year": 2026,
      "month": "2026-06",
      "title": "Large-scale study examines real coding-agent usage",
      "category": "Coding & Agents",
      "kind": "Research study",
      "access": "Published study",
      "organization": "Anthropic",
      "finding": "Anthropic studied about 400,000 Claude Code sessions to examine human-agent collaboration and expertise.",
      "possible": "Use the study to design code-agent oversight and human review roles.",
      "caution": "Session-based observational findings do not guarantee task success in another workflow.",
      "source": "https://www.anthropic.com/research/claude-code-expertise",
      "keywords": "coding adoption empirical",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-030",
      "date": "2026-06-09",
      "precision": "day",
      "year": 2026,
      "month": "2026-06",
      "title": "Live speech translation spans 70+ input languages",
      "category": "Accessibility & Language",
      "kind": "Product release",
      "impactMetric": "70+ spoken languages supported",
      "finding": "Google described continuous low-latency translation while preserving aspects of voice expression.",
      "possible": "More natural multilingual interactions can be prototyped in products.",
      "caution": "Noise, accents, mistakes and privacy remain significant concerns.",
      "source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-live-3-5-translate/",
      "organization": "Google",
      "access": "Product release",
      "keywords": "live speech translation spans 70+ input languages accessibility & language product release 70+ spoken languages supported",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "impact-038",
      "date": "2026-05-27",
      "precision": "day",
      "year": 2026,
      "month": "2026-05",
      "title": "Researchers experiment widely with AI but fewer use agents",
      "category": "Education & Learning",
      "kind": "Survey",
      "impactMetric": "81% chatbots vs 20% coding agents",
      "finding": "A survey of 1,260 social scientists found substantial chatbot use but lower agent use.",
      "possible": "Academic teams can target realistic training and workflow support.",
      "caution": "Self-reported uptake does not show causally improved research.",
      "source": "https://www.anthropic.com/research/coding-agents-social-sciences",
      "organization": "Anthropic",
      "access": "Survey",
      "keywords": "researchers experiment widely with ai but fewer use agents education & learning survey 81% chatbots vs 20% coding agents",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend"
    },
    {
      "id": "impact-015",
      "date": "2026-05-07",
      "precision": "day",
      "year": 2026,
      "month": "2026-05",
      "title": "AI-designed algorithms improve live infrastructure",
      "category": "Infrastructure & Energy",
      "kind": "Production deployment",
      "impactMetric": "20% lower write amplification in Google Spanner example",
      "finding": "Google described AlphaEvolve ideas integrated into TPU hardware and database heuristics.",
      "possible": "Algorithm search can measurably improve computing systems in production.",
      "caution": "Vendor-specific gains need separate testing before generalizing.",
      "source": "https://deepmind.google/blog/alphaevolve-impact/",
      "organization": "Google DeepMind",
      "access": "Production deployment",
      "keywords": "ai-designed algorithms improve live infrastructure infrastructure & energy production deployment 20% lower write amplification in google spanner example",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "adv-013",
      "date": "2026-05-07",
      "precision": "day",
      "year": 2026,
      "month": "2026-05",
      "title": "AlphaEvolve designs and improves algorithms across fields",
      "category": "Science",
      "kind": "Research milestone",
      "access": "Research-stage",
      "organization": "Google DeepMind",
      "finding": "AlphaEvolve research described algorithm improvements and deployments in some Google infrastructure.",
      "possible": "Explore algorithm-search assistance for narrowly testable optimization problems.",
      "caution": "Company case studies do not prove equivalent gains for every organization.",
      "source": "https://deepmind.google/blog/alphaevolve-impact/",
      "keywords": "algorithms math infrastructure optimization",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-032",
      "date": "2026-05-07",
      "precision": "day",
      "year": 2026,
      "month": "2026-05",
      "title": "Real-time voice models combine reasoning and translation",
      "category": "Accessibility & Language",
      "kind": "Developer platform release",
      "impactMetric": "Three specialized audio models",
      "finding": "OpenAI introduced new streaming voice reasoning, translation and transcription models.",
      "possible": "Developers can build complex voice interfaces with lower delay.",
      "caution": "Test speech accuracy, consent and costs before deployment.",
      "source": "https://openai.com/index/advancing-voice-intelligence-with-new-models-in-the-api/",
      "organization": "OpenAI",
      "access": "Developer platform release",
      "keywords": "real-time voice models combine reasoning and translation accessibility & language developer platform release three specialized audio models",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "impact-024",
      "date": "2026-05",
      "precision": "month",
      "year": 2026,
      "month": "2026-05",
      "title": "AI materials models generate candidate inorganic structures",
      "category": "Materials & Manufacturing",
      "kind": "Research capability",
      "impactMetric": "Generative materials design",
      "finding": "Microsoft Research documented MatterGen and MatterSim approaches for inverse materials design and simulation.",
      "possible": "Researchers can rank property-constrained candidate crystals faster.",
      "caution": "Candidates require synthesis and physical validation.",
      "source": "https://www.microsoft.com/en-us/research/project/materials/",
      "organization": "Microsoft Research",
      "access": "Research capability",
      "keywords": "ai materials models generate candidate inorganic structures materials & manufacturing research capability generative materials design",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "impact-036",
      "date": "2026-03-05",
      "precision": "day",
      "year": 2026,
      "month": "2026-03",
      "title": "Early labor research finds uneven AI displacement signals",
      "category": "Economy & Labor",
      "kind": "Observational study",
      "impactMetric": "No systematic unemployment rise detected",
      "finding": "Anthropic compared observed AI usage with occupation exposure and labor data.",
      "possible": "Researchers can monitor which tasks and jobs show evidence of disruption.",
      "caution": "No causal claim of job destruction or permanent safety is supported.",
      "source": "https://www.anthropic.com/research/labor-market-impacts",
      "organization": "Anthropic",
      "access": "Observational study",
      "keywords": "early labor research finds uneven ai displacement signals economy & labor observational study no systematic unemployment rise detected",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend"
    },
    {
      "id": "impact-012",
      "date": "2026-03-04",
      "precision": "day",
      "year": 2026,
      "month": "2026-03",
      "title": "Evo 2 DNA foundation model reaches Nature",
      "category": "Biology & Medicine",
      "kind": "Peer-reviewed research",
      "impactMetric": "Open genomics model",
      "finding": "Arc reported Evo 2's publication in Nature after its earlier preprint.",
      "possible": "Open biological model research gains peer-reviewed publication and new application studies.",
      "caution": "Not equivalent to a medically approved test.",
      "source": "https://arcinstitute.org/news/evo-2-one-year-later",
      "organization": "Arc Institute",
      "access": "Peer-reviewed research",
      "keywords": "evo 2 dna foundation model reaches nature biology & medicine peer-reviewed research open genomics model",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence"
    },
    {
      "id": "impact-017",
      "date": "2026-02-11",
      "precision": "day",
      "year": 2026,
      "month": "2026-02",
      "title": "AI contributes to original mathematical research",
      "category": "Mathematics & Algorithms",
      "kind": "Research collaboration",
      "impactMetric": "Multiple worked open questions and papers",
      "finding": "DeepMind described Aletheia work on mathematical conjectures with explicit human versus AI contribution levels.",
      "possible": "Researchers can test AI as a proof assistant and conjecture partner.",
      "caution": "The authors did not claim the highest-tier landmark advances.",
      "source": "https://deepmind.google/blog/accelerating-mathematical-and-scientific-discovery-with-gemini-deep-think/",
      "organization": "Google DeepMind",
      "access": "Research collaboration",
      "keywords": "ai contributes to original mathematical research mathematics & algorithms research collaboration multiple worked open questions and papers",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-012",
      "date": "2026-02-10",
      "precision": "day",
      "year": 2026,
      "month": "2026-02",
      "title": "Deep research gains connected and bounded sources",
      "category": "Research",
      "kind": "Feature update",
      "access": "Plan-dependent",
      "organization": "OpenAI",
      "finding": "Deep research added support for connected sources and trusted-site search restrictions.",
      "possible": "Run scoped research across approved internal records and nominated sources.",
      "caution": "Access varies by plan and connected source permissions.",
      "source": "https://openai.com/index/introducing-deep-research/",
      "keywords": "MCP connector search research",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-011",
      "date": "2026-02-05",
      "precision": "day",
      "year": 2026,
      "month": "2026-02",
      "title": "Codex takes on longer professional computer tasks",
      "category": "Coding & Agents",
      "kind": "Model release",
      "access": "Product-dependent",
      "organization": "OpenAI",
      "finding": "GPT-5.3-Codex extended agentic coding into longer research, tool use and execution tasks.",
      "possible": "Try multi-step code maintenance and review checkpoints rather than single-turn snippets.",
      "caution": "Model-provider claims require representative testing and can change with newer releases.",
      "source": "https://openai.com/index/introducing-gpt-5-3-codex/",
      "keywords": "coding computer professional",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-008",
      "date": "2026-02",
      "precision": "month",
      "year": 2026,
      "month": "2026-02",
      "title": "Robotaxi service scales past 400,000 weekly rides",
      "category": "Mobility & Transport",
      "kind": "Commercial deployment",
      "impactMetric": "400,000+ weekly rides",
      "finding": "Waymo reported substantial growth in driverless ride-hailing across major metro markets.",
      "possible": "More people can use fully driverless transport as an actual service.",
      "caution": "Availability remains geographically limited and depends on off-site operations.",
      "source": "https://waymo.com/blog/2026/02/waymo-raises-usd16-billion-investment-round/",
      "organization": "Waymo",
      "access": "Commercial deployment",
      "keywords": "robotaxi service scales past 400,000 weekly rides mobility & transport commercial deployment 400,000+ weekly rides",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "impact-028",
      "date": "2026-01-13",
      "precision": "day",
      "year": 2026,
      "month": "2026-01",
      "title": "Vertical AI-generated video targets mobile creators",
      "category": "Creative Media",
      "kind": "Product release",
      "impactMetric": "Portrait outputs with upscaling",
      "finding": "Google updated Veo 3.1 with portrait-oriented creation and improved control.",
      "possible": "Creators can produce short-form video variants more directly.",
      "caution": "Results may be misleading or visually inconsistent.",
      "source": "https://blog.google/innovation-and-ai/technology/ai/veo-3-1-ingredients-to-video/",
      "organization": "Google",
      "access": "Product release",
      "keywords": "vertical ai-generated video targets mobile creators creative media product release portrait outputs with upscaling",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "adv-010",
      "date": "2025-12-30",
      "precision": "day",
      "year": 2025,
      "month": "2025-12",
      "title": "Agent-native APIs become a production pattern",
      "category": "Infrastructure",
      "kind": "Industry retrospective",
      "access": "Contextual analysis",
      "organization": "OpenAI",
      "finding": "OpenAI's 2025 developer review described the shift from prompting toward agent tools, reasoning controls and multimodal APIs.",
      "possible": "Design small bounded automation pilots with evaluation, rollback and monitoring.",
      "caution": "An industry retrospective is not a new model launch or proof of ROI.",
      "source": "https://developers.openai.com/blog/openai-for-developers-2025",
      "keywords": "agents APIs automation",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-004",
      "date": "2025-12",
      "precision": "month",
      "year": 2025,
      "month": "2025-12",
      "title": "AI flood forecasting expands across countries",
      "category": "Climate & Weather",
      "kind": "Real-world deployment",
      "impactMetric": "2 billion people in covered risk areas",
      "finding": "Google Research reported flood prediction coverage in 150 countries for significant riverine floods.",
      "possible": "Emergency planners gain another early-warning signal for high-risk locations.",
      "caution": "Coverage does not mean every person receives an accurate personalized warning.",
      "source": "https://research.google/blog/google-research-2025-bolder-breakthroughs-bigger-impact/",
      "organization": "Google",
      "access": "Real-world deployment",
      "keywords": "ai flood forecasting expands across countries climate & weather real-world deployment 2 billion people in covered risk areas",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "impact-003",
      "date": "2025-12",
      "precision": "month",
      "year": 2025,
      "month": "2025-12",
      "title": "AI-powered monsoon forecasts reach Indian farmers",
      "category": "Food & Agriculture",
      "kind": "Real-world deployment",
      "impactMetric": "38 million farmers received forecasts",
      "finding": "Google and partners reported delivery of longer-range monsoon forecasting to farmers in India.",
      "possible": "Farmers and advisers can consider new predictive signals when planning crops and irrigation.",
      "caution": "Reach is not proof of better harvests for every farmer.",
      "source": "https://research.google/blog/google-research-2025-bolder-breakthroughs-bigger-impact/",
      "organization": "Google",
      "access": "Real-world deployment",
      "keywords": "ai-powered monsoon forecasts reach indian farmers food & agriculture real-world deployment 38 million farmers received forecasts",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "impact-005",
      "date": "2025-12",
      "precision": "month",
      "year": 2025,
      "month": "2025-12",
      "title": "FireSat identifies small fires missed by other satellites",
      "category": "Climate & Weather",
      "kind": "Field demonstration",
      "impactMetric": "Early small-fire detection",
      "finding": "Google Research described a FireSat sensor spotting a relatively small roadside fire that other space-based systems missed.",
      "possible": "Earlier detection may improve emergency response when systems scale.",
      "caution": "A pilot detection does not prove a full planned constellation is operational.",
      "source": "https://research.google/blog/google-research-2025-bolder-breakthroughs-bigger-impact/",
      "organization": "Google",
      "access": "Field demonstration",
      "keywords": "firesat identifies small fires missed by other satellites climate & weather field demonstration early small-fire detection",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "impact-007",
      "date": "2025-12",
      "precision": "month",
      "year": 2025,
      "month": "2025-12",
      "title": "Robotaxis serve millions of commercial passenger trips",
      "category": "Mobility & Transport",
      "kind": "Commercial deployment",
      "impactMetric": "More than 14 million trips reported for 2025",
      "finding": "Waymo's annual report described over 14 million rides during 2025.",
      "possible": "Driverless rides became ordinary paid transport in supported service areas.",
      "caution": "Operator-reported trips are not universal autonomy across all weather and roads.",
      "source": "https://waymo.com/blog/2026/02/waymo-raises-usd16-billion-investment-round/",
      "organization": "Waymo",
      "access": "Commercial deployment",
      "keywords": "robotaxis serve millions of commercial passenger trips mobility & transport commercial deployment more than 14 million trips reported for 2025",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "adv-009",
      "date": "2025-10-06",
      "precision": "day",
      "year": 2025,
      "month": "2025-10",
      "title": "Codex broadens engineering-team integrations",
      "category": "Coding & Agents",
      "kind": "Product release",
      "access": "Product available",
      "organization": "OpenAI",
      "finding": "Codex general availability included SDK and team-oriented workflow options.",
      "possible": "Connect code review and engineering workflows with auditable checkpoints.",
      "caution": "Integrations and platform entitlements evolve; verify current documentation.",
      "source": "https://openai.com/index/codex-now-generally-available/",
      "keywords": "development software sdk",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-027",
      "date": "2025-09-30",
      "precision": "day",
      "year": 2025,
      "month": "2025-09",
      "title": "AI video and dialogue generation become synchronized",
      "category": "Creative Media",
      "kind": "Historical product release",
      "impactMetric": "Sora 2 synchronized sound",
      "finding": "OpenAI launched Sora 2 with stronger motion control and synchronized audio.",
      "possible": "The release showed growing capacity for rapid audiovisual prototyping.",
      "caution": "Standalone Sora later closed in April 2026; this is a historical milestone, not current availability.",
      "source": "https://openai.com/index/sora-2/",
      "organization": "OpenAI",
      "access": "Historical product release",
      "keywords": "ai video and dialogue generation become synchronized creative media historical product release sora 2 synchronized sound",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-008",
      "date": "2025-09-29",
      "precision": "day",
      "year": 2025,
      "month": "2025-09",
      "title": "Claude Sonnet 4.5 expands coding and computer use",
      "category": "Coding & Agents",
      "kind": "Model release",
      "access": "Product available",
      "organization": "Anthropic",
      "finding": "Anthropic reported improvements in coding, agent work and computer-use capability.",
      "possible": "Pilot supervised coding and tool-driven tasks, checking output and permissions.",
      "caution": "Vendor-reported performance is not a universal independent ranking.",
      "source": "https://www.anthropic.com/news/claude-sonnet-4-5",
      "keywords": "coding agent computer use",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-033",
      "date": "2025-09-29",
      "precision": "day",
      "year": 2025,
      "month": "2025-09",
      "title": "Coding agents gain rollback and deeper tooling",
      "category": "Software & Work",
      "kind": "Product release",
      "impactMetric": "Checkpoints and VS Code integration",
      "finding": "Anthropic shipped coding-agent checkpoints and longer-context tooling.",
      "possible": "Engineers can delegate bounded changes with review and rollback.",
      "caution": "Provider performance claims require task-specific evaluation.",
      "source": "https://www.anthropic.com/news/claude-sonnet-4-5",
      "organization": "Anthropic",
      "access": "Product release",
      "keywords": "coding agents gain rollback and deeper tooling software & work product release checkpoints and vs code integration",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "impact-031",
      "date": "2025-08-28",
      "precision": "day",
      "year": 2025,
      "month": "2025-08",
      "title": "Real-time voice agents gain tools and calling",
      "category": "Accessibility & Language",
      "kind": "Developer platform release",
      "impactMetric": "SIP calling and external tools",
      "finding": "OpenAI made Realtime API generally available with richer speech, phone and tool capabilities.",
      "possible": "Businesses can prototype supervised voice workflows with authorized tools.",
      "caution": "Human escalation, disclosure and account permissions are essential.",
      "source": "https://openai.com/index/introducing-gpt-realtime/",
      "organization": "OpenAI",
      "access": "Developer platform release",
      "keywords": "real-time voice agents gain tools and calling accessibility & language developer platform release sip calling and external tools",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "impact-029",
      "date": "2025-08-26",
      "precision": "day",
      "year": 2025,
      "month": "2025-08",
      "title": "Live language translation reaches phone conversations",
      "category": "Accessibility & Language",
      "kind": "Product release",
      "impactMetric": "Live spoken translation",
      "finding": "Google Translate added live multilingual audio conversations.",
      "possible": "Travelers and multilingual groups can reduce communication friction.",
      "caution": "Translation nuance and reliability vary by language.",
      "source": "https://blog.google/products-and-platforms/products/translate/language-learning-live-translate/",
      "organization": "Google",
      "access": "Product release",
      "keywords": "live language translation reaches phone conversations accessibility & language product release live spoken translation",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "adv-007",
      "date": "2025-08-07",
      "precision": "day",
      "year": 2025,
      "month": "2025-08",
      "title": "GPT-5 introduces built-in reasoning routing",
      "category": "Models",
      "kind": "Model release",
      "access": "Product available",
      "organization": "OpenAI",
      "finding": "GPT-5 combined faster responses and longer deliberation under one model family.",
      "possible": "Compare fast and deliberate reasoning on representative professional tasks.",
      "caution": "This is a historical release, not a claim that GPT-5 remains the latest model.",
      "source": "https://openai.com/index/introducing-gpt-5/",
      "keywords": "reasoning model",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-018",
      "date": "2025-08-05",
      "precision": "day",
      "year": 2025,
      "month": "2025-08",
      "title": "Text prompts create navigable interactive virtual worlds",
      "category": "Creative Media",
      "kind": "Research demonstration",
      "impactMetric": "720p at 24 fps for minutes",
      "finding": "Genie 3 demonstrated short-lived interactive worlds navigable in real time.",
      "possible": "Research teams can prototype simulations and agent training environments.",
      "caution": "Generated environments do not guarantee real-world physical fidelity.",
      "source": "https://deepmind.google/blog/genie-3-a-new-frontier-for-world-models/",
      "organization": "Google DeepMind",
      "access": "Research demonstration",
      "keywords": "text prompts create navigable interactive virtual worlds creative media research demonstration 720p at 24 fps for minutes",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "impact-016",
      "date": "2025-07-21",
      "precision": "day",
      "year": 2025,
      "month": "2025-07",
      "title": "AI solutions reach gold-medal math standard",
      "category": "Mathematics & Algorithms",
      "kind": "Externally graded benchmark",
      "impactMetric": "35 of 42 IMO points",
      "finding": "Gemini Deep Think solved five International Mathematical Olympiad problems under official grading.",
      "possible": "Advanced mathematical reasoning has crossed a notable competition milestone.",
      "caution": "Competition performance does not prove general mathematical research reliability.",
      "source": "https://deepmind.google/blog/advanced-version-of-gemini-with-deep-think-officially-achieves-gold-medal-standard-at-the-international-mathematical-olympiad/",
      "organization": "Google DeepMind",
      "access": "Externally graded benchmark",
      "keywords": "ai solutions reach gold-medal math standard mathematics & algorithms externally graded benchmark 35 of 42 imo points",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence"
    },
    {
      "id": "adv-006",
      "date": "2025-07-17",
      "precision": "day",
      "year": 2025,
      "month": "2025-07",
      "title": "ChatGPT agent combines research and browser actions",
      "category": "Agents",
      "kind": "Product release",
      "access": "Plan-dependent",
      "organization": "OpenAI",
      "finding": "A combined agent could research, navigate websites and work across tools during a supervised task.",
      "possible": "Delegate multi-step research and drafting with permission boundaries and checkpoints.",
      "caution": "Product controls, plan access and reliability change over time.",
      "source": "https://openai.com/index/introducing-chatgpt-agent/",
      "keywords": "agent browsers tasks research",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-034",
      "date": "2025-07-10",
      "precision": "day",
      "year": 2025,
      "month": "2025-07",
      "title": "AI coding makes some expert developers slower",
      "category": "Software & Work",
      "kind": "Independent controlled study",
      "impactMetric": "19% longer completion time",
      "finding": "METR randomized 246 tasks across 16 experienced developers and found early-2025 AI tools increased task time.",
      "possible": "Organizations should measure real workflow speed rather than believe every benchmark.",
      "caution": "A small sample and early tool versions limit generalization.",
      "source": "https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/",
      "organization": "METR",
      "access": "Independent controlled study",
      "keywords": "ai coding makes some expert developers slower software & work independent controlled study 19% longer completion time",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence"
    },
    {
      "id": "impact-026",
      "date": "2025-07-10",
      "precision": "day",
      "year": 2025,
      "month": "2025-07",
      "title": "Still photos turn into short videos with audio",
      "category": "Creative Media",
      "kind": "Product expansion",
      "impactMetric": "Eight-second photo-to-video clips",
      "finding": "Google expanded Veo-powered creation to animate images with sound.",
      "possible": "Creators can animate rights-cleared photographs and mockups.",
      "caution": "Synthetic footage must not be represented as recorded events.",
      "source": "https://blog.google/products-and-platforms/products/gemini/photo-to-video/",
      "organization": "Google",
      "access": "Product expansion",
      "keywords": "still photos turn into short videos with audio creative media product expansion eight-second photo-to-video clips",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "impact-002",
      "date": "2025-07-01",
      "precision": "day",
      "year": 2025,
      "month": "2025-07",
      "title": "Operational AI weather ensembles arrive",
      "category": "Climate & Weather",
      "kind": "Operational deployment",
      "impactMetric": "51 ensemble forecasts",
      "finding": "ECMWF deployed an AI-driven probabilistic forecast ensemble in parallel with conventional methods.",
      "possible": "Meteorologists can compare model uncertainty and multiple potential outcomes for decision making.",
      "caution": "Lower resolution and variable-dependent performance require conventional cross-checks.",
      "source": "https://www.ecmwf.int/en/about/media-centre/news/2025/ecmwfs-ensemble-ai-forecasts-become-operational",
      "organization": "ECMWF",
      "access": "Operational deployment",
      "keywords": "operational ai weather ensembles arrive climate & weather operational deployment 51 ensemble forecasts",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "impact-022",
      "date": "2025-06-25",
      "precision": "day",
      "year": 2025,
      "month": "2025-06",
      "title": "AlphaGenome begins decoding variant consequences",
      "category": "Biology & Medicine",
      "kind": "Research release",
      "impactMetric": "Unified DNA variant-effect prediction",
      "finding": "DeepMind released AlphaGenome to predict multiple molecular properties of genetic variants.",
      "possible": "Researchers can better prioritize questions about gene regulation.",
      "caution": "These are modeled effects, not clinical diagnoses.",
      "source": "https://deepmind.google/blog/alphagenome-ai-for-better-understanding-the-genome/",
      "organization": "Google DeepMind",
      "access": "Research release",
      "keywords": "alphagenome begins decoding variant consequences biology & medicine research release unified dna variant-effect prediction",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-005",
      "date": "2025-06-25",
      "precision": "day",
      "year": 2025,
      "month": "2025-06",
      "title": "AlphaGenome predicts effects of DNA variants",
      "category": "Science",
      "kind": "Research release",
      "access": "Research API",
      "organization": "Google DeepMind",
      "finding": "AlphaGenome introduced predictions about regulatory effects of DNA sequence variants.",
      "possible": "Prioritize experimental questions in genomics, subject to specialist assessment.",
      "caution": "Predictions are not clinical diagnoses or experimentally confirmed biological effects.",
      "source": "https://deepmind.google/blog/alphagenome-ai-for-better-understanding-the-genome/",
      "keywords": "biology genetics genomics dna",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "adv-004",
      "date": "2025-05-16",
      "precision": "day",
      "year": 2025,
      "month": "2025-05",
      "title": "Cloud coding agents work on repository tasks",
      "category": "Coding & Agents",
      "kind": "Product preview",
      "access": "Product-dependent",
      "organization": "OpenAI",
      "finding": "Codex research preview could write code, investigate bugs and propose pull requests in isolated environments.",
      "possible": "Delegate bounded test generation and code changes while reviewing diffs and CI outcomes.",
      "caution": "Original preview capabilities and availability differ from today's Codex versions.",
      "source": "https://openai.com/index/introducing-codex/",
      "keywords": "code test repository software engineering",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-014",
      "date": "2025-05-14",
      "precision": "day",
      "year": 2025,
      "month": "2025-05",
      "title": "AlphaEvolve searches automatically for better algorithms",
      "category": "Mathematics & Algorithms",
      "kind": "Research demonstration",
      "impactMetric": "Automatic propose-and-evaluate algorithm search",
      "finding": "DeepMind launched AlphaEvolve for discovering computational algorithms and mathematics improvements.",
      "possible": "Teams can test candidate optimization strategies where success has a measurable objective.",
      "caution": "Only reliably checkable problems suit this approach well.",
      "source": "https://deepmind.google/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/",
      "organization": "Google DeepMind",
      "access": "Research demonstration",
      "keywords": "alphaevolve searches automatically for better algorithms mathematics & algorithms research demonstration automatic propose-and-evaluate algorithm search",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "impact-019",
      "date": "2025-05",
      "precision": "month",
      "year": 2025,
      "month": "2025-05",
      "title": "Aurora predicts varied Earth-system phenomena",
      "category": "Climate & Weather",
      "kind": "Peer-reviewed research",
      "impactMetric": "More than one million hours of geophysical data",
      "finding": "Microsoft researchers published Aurora, an Earth-system foundation model tested on several environmental tasks.",
      "possible": "Scientists can reuse a pretrained model across weather and environmental studies.",
      "caution": "Research comparisons do not replace official forecasts.",
      "source": "https://www.microsoft.com/en-us/research/publication/aurora-a-foundation-model-for-the-earth-system/",
      "organization": "Microsoft Research",
      "access": "Peer-reviewed research",
      "keywords": "aurora predicts varied earth-system phenomena climate & weather peer-reviewed research more than one million hours of geophysical data",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence"
    },
    {
      "id": "impact-annual-046",
      "date": "2025-04-10",
      "precision": "day",
      "year": 2025,
      "month": "2025-04",
      "title": "Electricity demand projections reveal AI capacity constraints",
      "category": "Infrastructure & Energy",
      "kind": "Scenario forecast",
      "impactMetric": "945 TWh global data-center demand by 2030",
      "finding": "IEA projects more than doubling demand with AI a major driver.",
      "possible": "Utilities need scenario-based capacity planning and permitting.",
      "caution": "A projection is not a realized 2030 outcome.",
      "source": "https://www.iea.org/reports/energy-and-ai/executive-summary",
      "organization": "International Energy Agency",
      "access": "Scenario forecast",
      "keywords": "electricity demand projections reveal ai capacity constraints infrastructure & energy 945 twh global data-center demand by 2030",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Source publication date."
    },
    {
      "id": "impact-annual-045",
      "date": "2025-04-10",
      "precision": "day",
      "year": 2025,
      "month": "2025-04",
      "title": "IEA measures the energy footprint behind the AI boom",
      "category": "Infrastructure & Energy",
      "kind": "Independent energy analysis",
      "impactMetric": "415 TWh global data-center electricity in 2024",
      "finding": "IEA reports data-center electricity at roughly 1.5% of world consumption.",
      "possible": "Grid planners can see the actual baseline for digital infrastructure demand.",
      "caution": "Not all electricity in these data centers is used by AI.",
      "source": "https://www.iea.org/reports/energy-and-ai/executive-summary",
      "organization": "International Energy Agency",
      "access": "Independent energy analysis",
      "keywords": "iea measures the energy footprint behind the ai boom infrastructure & energy 415 twh global data-center electricity in 2024",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence",
      "sourceContext": "Source publication date."
    },
    {
      "id": "impact-025",
      "date": "2025-03-25",
      "precision": "day",
      "year": 2025,
      "month": "2025-03",
      "title": "AI image generation becomes iterative and text-aware",
      "category": "Creative Media",
      "kind": "Product release",
      "impactMetric": "Multimodal image editing",
      "finding": "OpenAI introduced integrated image generation with improved text placement and visual editing.",
      "possible": "Creative teams can rapidly prototype diagrams and marketing concepts.",
      "caution": "Generated images can contain inaccurate details and intellectual-property risks.",
      "source": "https://openai.com/index/introducing-4o-image-generation/",
      "organization": "OpenAI",
      "access": "Product release",
      "keywords": "ai image generation becomes iterative and text-aware creative media product release multimodal image editing",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Product capability"
    },
    {
      "id": "adv-003",
      "date": "2025-03-12",
      "precision": "day",
      "year": 2025,
      "month": "2025-03",
      "title": "Gemini Robotics links multimodal AI to physical action",
      "category": "Robotics",
      "kind": "Research demonstration",
      "access": "Research-stage",
      "organization": "Google DeepMind",
      "finding": "Gemini Robotics and Robotics-ER linked visual and language reasoning with robot control in demonstrations.",
      "possible": "Study emerging robot-assistance tasks in handling, sorting and manipulation.",
      "caution": "Demonstration does not prove deployment safety or availability in ordinary workplaces.",
      "source": "https://deepmind.google/blog/gemini-robotics-brings-ai-into-the-physical-world/",
      "keywords": "robot manipulation physical dexterity",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-001",
      "date": "2025-02-25",
      "precision": "day",
      "year": 2025,
      "month": "2025-02",
      "title": "AI weather forecasts become operational",
      "category": "Climate & Weather",
      "kind": "Operational deployment",
      "impactMetric": "Up to 1,000× lower energy per forecast",
      "finding": "ECMWF launched the Artificial Intelligence Forecasting System alongside conventional forecasts and reported stronger performance on several metrics.",
      "possible": "Weather agencies can run complementary global forecasts at significantly lower computational cost.",
      "caution": "AIFS complements physics models; resolution and regional skill differ.",
      "source": "https://www.ecmwf.int/en/about/media-centre/news/2025/ecmwfs-ai-forecasts-become-operational",
      "organization": "ECMWF",
      "access": "Operational deployment",
      "keywords": "ai weather forecasts become operational climate & weather operational deployment up to 1,000× lower energy per forecast",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Real-world deployment"
    },
    {
      "id": "impact-010",
      "date": "2025-02-19",
      "precision": "day",
      "year": 2025,
      "month": "2025-02",
      "title": "Evo 2 learns from genetic code across domains of life",
      "category": "Biology & Medicine",
      "kind": "Research release",
      "impactMetric": "9.3 trillion DNA tokens",
      "finding": "Arc Institute launched a large open DNA language model for biological sequence prediction and generation.",
      "possible": "Scientists can generate and rank sequence hypotheses computationally.",
      "caution": "Biosafety and wet-lab confirmation remain crucial.",
      "source": "https://arcinstitute.org/news/evo2",
      "organization": "Arc Institute",
      "access": "Research release",
      "keywords": "evo 2 learns from genetic code across domains of life biology & medicine research release 9.3 trillion dna tokens",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "impact-035",
      "date": "2025-02-10",
      "precision": "day",
      "year": 2025,
      "month": "2025-02",
      "title": "Usage evidence shows AI penetrating many occupations",
      "category": "Economy & Labor",
      "kind": "Observational study",
      "impactMetric": "36% of jobs used AI on at least a quarter of tasks",
      "finding": "Anthropic analyzed anonymized Claude usage across occupational tasks.",
      "possible": "Businesses can focus on task redesign rather than declaring whole occupations obsolete.",
      "caution": "Single-platform use is not an unbiased cross-economy measure.",
      "source": "https://www.anthropic.com/research/the-anthropic-economic-index",
      "organization": "Anthropic",
      "access": "Observational study",
      "keywords": "usage evidence shows ai penetrating many occupations economy & labor observational study 36% of jobs used ai on at least a quarter of tasks",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend"
    },
    {
      "id": "adv-002",
      "date": "2025-02-02",
      "precision": "day",
      "year": 2025,
      "month": "2025-02",
      "title": "Deep research assembles multi-source reports",
      "category": "Research",
      "kind": "Product release",
      "access": "Available with plan limits",
      "organization": "OpenAI",
      "finding": "Deep research introduced an agentic workflow that searches and synthesizes external material into sourced reports.",
      "possible": "Ask for comparative research with traceable citations, then verify the important claims.",
      "caution": "Research outputs can miss sources and include citation errors.",
      "source": "https://openai.com/index/introducing-deep-research/",
      "keywords": "research citations sourcing synthesis",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Product capability",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-011",
      "date": "2025-02",
      "precision": "month",
      "year": 2025,
      "month": "2025-02",
      "title": "Virtual-cell datasets begin supporting predictive biology",
      "category": "Biology & Medicine",
      "kind": "Open research dataset",
      "impactMetric": "More than 600 million cells in expanding atlas",
      "finding": "Arc's Virtual Cell Atlas combines observations and perturbation records for training biological models.",
      "possible": "Researchers can study how cellular systems react to interventions.",
      "caution": "Dataset quantity does not establish causal validity or experimental generalization.",
      "source": "https://arcinstitute.org/tools/virtualcellatlas",
      "organization": "Arc Institute",
      "access": "Open research dataset",
      "keywords": "virtual-cell datasets begin supporting predictive biology biology & medicine open research dataset more than 600 million cells in expanding atlas",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration"
    },
    {
      "id": "adv-001",
      "date": "2025-01-23",
      "precision": "day",
      "year": 2025,
      "month": "2025-01",
      "title": "Browser agents begin performing web tasks",
      "category": "Agents",
      "kind": "Research preview",
      "access": "Limited access",
      "organization": "OpenAI",
      "finding": "Operator showed an AI agent using a browser to click, type and navigate websites.",
      "possible": "Experiment with supervised browser workflows; require explicit confirmations for purchases and sensitive actions.",
      "caution": "Standalone Operator later moved into ChatGPT agent; historical launch, not a currently separate product.",
      "source": "https://openai.com/index/introducing-operator/",
      "keywords": "browser automation research operator",
      "reviewed": "2026-11-07",
      "impactMetric": "",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Original dated milestone; publication and access terms may change."
    },
    {
      "id": "impact-annual-029",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Academia remains central to scientific AI infrastructure",
      "category": "Science & Discovery",
      "kind": "Research landscape synthesis",
      "impactMetric": "Academic and public-sector labs produce many science models",
      "finding": "Stanford finds scientific AI model development differs from commercial frontier-model concentration.",
      "possible": "University and public research partnerships remain important.",
      "caution": "Patterns vary widely between different scientific disciplines.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Research landscape synthesis",
      "keywords": "academia remains central to scientific ai infrastructure science & discovery academic and public-sector labs produce many science models",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-001",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI becomes a mass-population technology faster than the internet",
      "category": "Economy & Labor",
      "kind": "Population survey synthesis",
      "impactMetric": "53% adoption in three years",
      "finding": "Stanford reports unusually rapid public adoption of generative AI across studied countries.",
      "possible": "Governments and employers must adapt faster than earlier technology transitions.",
      "caution": "Country variation is wide; adoption is not proof of productivity.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/economy",
      "organization": "Stanford HAI",
      "access": "Population survey synthesis",
      "keywords": "ai becomes a mass-population technology faster than the internet economy & labor 53% adoption in three years",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-011",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI begins scoring strongly on select professional evaluations",
      "category": "Models & Benchmarks",
      "kind": "Professional benchmark synthesis",
      "impactMetric": "Roughly 60%–90% on selected tax, legal and finance evaluations",
      "finding": "Stanford surveys results showing models can answer some domain exams impressively.",
      "possible": "AI can assist with first drafts of regulated work under qualified review.",
      "caution": "Benchmark outcomes cannot substitute for professional liability and actual case quality.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Professional benchmark synthesis",
      "keywords": "ai begins scoring strongly on select professional evaluations models & benchmarks roughly 60%–90% on selected tax, legal and finance evaluations",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-043",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI computing power demands concentrate in infrastructure hubs",
      "category": "Infrastructure & Energy",
      "kind": "Infrastructure data",
      "impactMetric": "US hosts more than 5,400 data centers",
      "finding": "Stanford describes concentrated data-center geography and key chip supply bottlenecks.",
      "possible": "Electricity, cooling, land and hardware capacity now constrain expansion.",
      "caution": "The number includes non-AI data centers.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report",
      "organization": "Stanford HAI",
      "access": "Infrastructure data",
      "keywords": "ai computing power demands concentrate in infrastructure hubs infrastructure & energy us hosts more than 5,400 data centers",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-033",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI education becomes a national curriculum requirement in some countries",
      "category": "Education & Learning",
      "kind": "Public policy evidence",
      "impactMetric": "China and UAE 2025–26 mandates",
      "finding": "Stanford reports national-level AI education requirements entering school systems.",
      "possible": "AI literacy is becoming a public education objective.",
      "caution": "Mandates do not guarantee access or learning outcomes.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/education",
      "organization": "Stanford HAI",
      "access": "Public policy evidence",
      "keywords": "ai education becomes a national curriculum requirement in some countries education & learning china and uae 2025–26 mandates",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-003",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI investment more than doubles",
      "category": "Economy & Labor",
      "kind": "Investment data",
      "impactMetric": "About $581.7 billion total corporate AI investment",
      "finding": "Stanford reports exceptional 2025 global spending, with private funding and mergers driving growth.",
      "possible": "Capital allocation, research priorities and chip demand change at scale.",
      "caution": "Investment is not equivalent to economic returns.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/economy",
      "organization": "Stanford HAI",
      "access": "Investment data",
      "keywords": "ai investment more than doubles economy & labor about $581.7 billion total corporate ai investment",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-040",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI sovereignty drives government infrastructure investment",
      "category": "Infrastructure & Energy",
      "kind": "Policy research",
      "impactMetric": "44 public AI clusters in Europe and Central Asia",
      "finding": "Stanford reports state-backed compute expansion and uneven global capacity distribution.",
      "possible": "Access to computing becomes a national strategy consideration.",
      "caution": "Cluster counts do not equal comparable useful compute power.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/policy-and-governance",
      "organization": "Stanford HAI",
      "access": "Policy research",
      "keywords": "ai sovereignty drives government infrastructure investment infrastructure & energy 44 public ai clusters in europe and central asia",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-008",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI videos begin exhibiting surprising physical patterns",
      "category": "Creative Media",
      "kind": "Benchmark synthesis",
      "impactMetric": "18,000+ AI videos examined in a study",
      "finding": "Stanford reviews work finding some video models demonstrate zero-shot physical-pattern reasoning.",
      "possible": "Generated environments may become useful in simulation and training.",
      "caution": "Physical plausibility is not reliable simulation of reality.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "ai videos begin exhibiting surprising physical patterns creative media 18,000+ ai videos examined in a study",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-019",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI-generated answers dominate many health searches",
      "category": "Search & Information",
      "kind": "Search prevalence research",
      "impactMetric": "84%–92% of health queries showed AI summaries in studied samples",
      "finding": "Stanford reviews studies of generated search summaries appearing before traditional results.",
      "possible": "Health information publishers and patients now encounter machine summaries as a primary interface.",
      "caution": "Prevalence is not accuracy or medical advice quality.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Search prevalence research",
      "keywords": "ai-generated answers dominate many health searches search & information 84%–92% of health queries showed ai summaries in studied samples",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-027",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "AI-written scientific paper reaches workshop acceptance",
      "category": "Science & Discovery",
      "kind": "Research demonstration",
      "impactMetric": "One AI-generated paper accepted at an ICLR workshop",
      "finding": "Stanford describes AI Scientist-v2 reaching a workshop peer-review milestone.",
      "possible": "AI can automate fragments of idea generation, experimentation and paper writing.",
      "caution": "One accepted workshop paper is not evidence of reliable autonomous science.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Research demonstration",
      "keywords": "ai-written scientific paper reaches workshop acceptance science & discovery one ai-generated paper accepted at an iclr workshop",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Research demonstration",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-025",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Astronomy gains foundation models across giant sky surveys",
      "category": "Science & Discovery",
      "kind": "Research infrastructure",
      "impactMetric": "More than 200 million celestial objects in one model training set",
      "finding": "Stanford describes AION-1 and associated astronomical datasets and benchmarks.",
      "possible": "Astronomers can explore cross-survey signal retrieval and classification.",
      "caution": "Training scale does not establish discovery quality.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Research infrastructure",
      "keywords": "astronomy gains foundation models across giant sky surveys science & discovery more than 200 million celestial objects in one model training set",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-028",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Bioinformatics agents miss most real-world analysis tasks",
      "category": "Science & Discovery",
      "kind": "Benchmark synthesis",
      "impactMetric": "Around 17% accuracy in BixBench",
      "finding": "Stanford highlights major limitations in automated bioinformatics tasks.",
      "possible": "Researchers can identify workflows where verification remains indispensable.",
      "caution": "AI can confidently generate wrong analysis outputs.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "bioinformatics agents miss most real-world analysis tasks science & discovery around 17% accuracy in bixbench",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-022",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Chemistry models exceed average expert test scores",
      "category": "Science & Discovery",
      "kind": "Benchmark synthesis",
      "impactMetric": "2,700+ ChemBench questions",
      "finding": "Stanford highlights chemistry-question performance reaching above average expert results.",
      "possible": "AI can accelerate scientific question triage and literature interrogation.",
      "caution": "Strong exam performance does not imply experimental replicability.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "chemistry models exceed average expert test scores science & discovery 2,700+ chembench questions",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-010",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Chinese and US models compete near the same frontier",
      "category": "Models & Benchmarks",
      "kind": "Benchmark synthesis",
      "impactMetric": "Top performance gap falls to single digits",
      "finding": "Stanford documents a narrowing capability gap between US and Chinese providers.",
      "possible": "Organizations can assess more model alternatives and regional options.",
      "caution": "Arena-style ratings are a particular benchmark, not universal intelligence.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "chinese and us models compete near the same frontier models & benchmarks top performance gap falls to single digits",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-012",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Clinical AI scribes reach hospital-scale use",
      "category": "Biology & Medicine",
      "kind": "Operational adoption reports",
      "impactMetric": "Up to 83% reduction in note-writing time in cases reviewed",
      "finding": "Stanford describes large health-system use of ambient documentation tools.",
      "possible": "Clinicians can spend less time on routine paperwork if edits are accurately reviewed.",
      "caution": "Reported benefits vary and note inaccuracies may harm care.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Operational adoption reports",
      "keywords": "clinical ai scribes reach hospital-scale use biology & medicine up to 83% reduction in note-writing time in cases reviewed",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-015",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Clinical proof lags behind AI device deployment",
      "category": "Biology & Medicine",
      "kind": "Evidence review",
      "impactMetric": "2.4% of reviewed supporting studies had randomized trials",
      "finding": "Stanford highlights gaps in prospective evidence supporting many medical AI authorizations.",
      "possible": "Health systems can require stronger patient-level evaluation before adoption.",
      "caution": "Reported fraction is conditional on the reviewed studies, not all devices.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Evidence review",
      "keywords": "clinical proof lags behind ai device deployment biology & medicine 2.4% of reviewed supporting studies had randomized trials",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-005",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Computer-use agents make enormous benchmark gains",
      "category": "Models & Benchmarks",
      "kind": "Benchmark synthesis",
      "impactMetric": "About 12% to 66.3% success on OSWorld",
      "finding": "Stanford tracks major improvements in AI agents performing cross-application computer tasks.",
      "possible": "Computer agents can tackle more supervised office and browsing workflows.",
      "caution": "They still fail about one in three structured benchmark attempts.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "computer-use agents make enormous benchmark gains models & benchmarks about 12% to 66.3% success on osworld",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-014",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Diagnostic agents excel on complex simulated cases",
      "category": "Biology & Medicine",
      "kind": "Clinical simulation",
      "impactMetric": "Up to 85.5% accuracy on a selected case benchmark",
      "finding": "Stanford examines a multi-agent medical diagnostic system evaluated on published clinical vignettes.",
      "possible": "Research can test multi-agent diagnostic support under supervision.",
      "caution": "Unaided-physician comparisons and simulated cases do not prove superior real-world care.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Clinical simulation",
      "keywords": "diagnostic agents excel on complex simulated cases biology & medicine up to 85.5% accuracy on a selected case benchmark",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-020",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Digital-twin diabetes intervention shows early results",
      "category": "Biology & Medicine",
      "kind": "Clinical trial synthesis",
      "impactMetric": "150-participant randomized study",
      "finding": "Stanford describes an early intervention trial reporting improvements in blood-sugar control.",
      "possible": "Personalized digital-twin interventions warrant larger controlled trials.",
      "caution": "The finding cannot be extrapolated to all conditions or patients.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Clinical trial synthesis",
      "keywords": "digital-twin diabetes intervention shows early results biology & medicine 150-participant randomized study",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-026",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "End-to-end learned weather pipelines emerge",
      "category": "Climate & Weather",
      "kind": "Research synthesis",
      "impactMetric": "Aardvark integrated forecasting research",
      "finding": "Stanford reports research toward replacing complicated prediction-pipeline stages with AI.",
      "possible": "Lower-cost forecasting architectures become viable research directions.",
      "caution": "Operational meteorology still requires verification and contingency planning.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Research synthesis",
      "keywords": "end-to-end learned weather pipelines emerge climate & weather aardvark integrated forecasting research",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-039",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "English-language AI quality masks global dialect failures",
      "category": "Accessibility & Language",
      "kind": "Multilingual benchmark",
      "impactMetric": "Large performance gaps for regional dialects",
      "finding": "Stanford documents weaker results in some non-English and regional language contexts.",
      "possible": "Organizations should test locally relevant language varieties directly.",
      "caution": "Strong English benchmarks do not guarantee multilingual reliability.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai",
      "organization": "Stanford HAI",
      "access": "Multilingual benchmark",
      "keywords": "english-language ai quality masks global dialect failures accessibility & language large performance gaps for regional dialects",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-004",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Estimated consumer benefits of generative AI rise",
      "category": "Economy & Labor",
      "kind": "Economic modeling",
      "impactMetric": "$172 billion annual US consumer surplus estimated by early 2026",
      "finding": "Stanford describes rapid growth in estimated value consumers derive from low-cost AI tools.",
      "possible": "People are gaining utility outside paid enterprise software.",
      "caution": "Consumer surplus is an estimate, not direct revenue or wage gains.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/economy",
      "organization": "Stanford HAI",
      "access": "Economic modeling",
      "keywords": "estimated consumer benefits of generative ai rise economy & labor $172 billion annual us consumer surplus estimated by early 2026",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-009",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Familiar AI benchmarks begin to saturate",
      "category": "Models & Benchmarks",
      "kind": "Benchmark validity study",
      "impactMetric": "As many as 42% invalid test questions in one reviewed benchmark",
      "finding": "Stanford documents benchmark contamination and quality issues.",
      "possible": "Businesses need independent tests on actual tasks rather than buying by a leaderboard rank.",
      "caution": "A score jump does not necessarily mean better deployment performance.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Benchmark validity study",
      "keywords": "familiar ai benchmarks begin to saturate models & benchmarks as many as 42% invalid test questions in one reviewed benchmark",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-044",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Frontier training compute grows at an extraordinary pace",
      "category": "Infrastructure & Energy",
      "kind": "Compute trend synthesis",
      "impactMetric": "Historical estimate around 3.3× annually",
      "finding": "Stanford reports rapid growth in training compute since 2022.",
      "possible": "Infrastructure concentration may shape who can compete at the frontier.",
      "caution": "Historical trends are not guaranteed to persist.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report",
      "organization": "Stanford HAI",
      "access": "Compute trend synthesis",
      "keywords": "frontier training compute grows at an extraordinary pace infrastructure & energy historical estimate around 3.3× annually",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-035",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Hallucination remains common under certain tests",
      "category": "Safety, Trust & Policy",
      "kind": "Benchmark synthesis",
      "impactMetric": "22%–94% across models on one test",
      "finding": "Stanford documents huge variation in model behavior when handling false user-supplied beliefs.",
      "possible": "High-stakes AI outputs need independent verification.",
      "caution": "One adversarial evaluation cannot summarize general factuality.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "hallucination remains common under certain tests safety, trust & policy 22%–94% across models on one test",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-006",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Household robots remain much less capable than demos suggest",
      "category": "Robotics",
      "kind": "Benchmark synthesis",
      "impactMetric": "About 12% real household task success",
      "finding": "Stanford contrasts strong robotics simulation performance with poor results in real homes.",
      "possible": "Practical physical automation should start in predictable environments.",
      "caution": "Viral robotics clips do not establish general household autonomy.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "household robots remain much less capable than demos suggest robotics about 12% real household task success",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-036",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Model-maker transparency declines in an index",
      "category": "Safety, Trust & Policy",
      "kind": "Transparency index",
      "impactMetric": "Average score declines to 40",
      "finding": "Stanford reports worsening disclosure of model data, training and post-deployment effects.",
      "possible": "External assurance becomes harder as systems scale.",
      "caution": "Index scores depend on measurement criteria.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai",
      "organization": "Stanford HAI",
      "access": "Transparency index",
      "keywords": "model-maker transparency declines in an index safety, trust & policy average score declines to 40",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-030",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Most high-school and college students use AI for schoolwork",
      "category": "Education & Learning",
      "kind": "Education survey",
      "impactMetric": "More than 80% reported use",
      "finding": "Stanford describes widespread AI use for homework, writing and research.",
      "possible": "Education systems need source literacy, assessment redesign and AI instruction.",
      "caution": "Use does not prove learning gains.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/education",
      "organization": "Stanford HAI",
      "access": "Education survey",
      "keywords": "most high-school and college students use ai for schoolwork education & learning more than 80% reported use",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-018",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Much clinical-AI testing still relies on simulated patients",
      "category": "Biology & Medicine",
      "kind": "Evidence review",
      "impactMetric": "Nearly half of clinical AI studies use simulated scenarios",
      "finding": "Stanford identifies a gap between published diagnostic demonstrations and patient-level trials.",
      "possible": "Clinical developers need prospective evaluations that reflect real clinics.",
      "caution": "Excellent simulation accuracy is not a patient outcome.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Evidence review",
      "keywords": "much clinical-ai testing still relies on simulated patients biology & medicine nearly half of clinical ai studies use simulated scenarios",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-041",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "National AI strategies spread across emerging economies",
      "category": "Safety, Trust & Policy",
      "kind": "Policy research",
      "impactMetric": "New national planning across more regions",
      "finding": "Stanford documents government strategy adoption outside early AI powers.",
      "possible": "AI affects education, national infrastructure and economic policy worldwide.",
      "caution": "Policy plans differ from executed projects.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/policy-and-governance",
      "organization": "Stanford HAI",
      "access": "Policy research",
      "keywords": "national ai strategies spread across emerging economies safety, trust & policy new national planning across more regions",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-007",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Olympiad reasoning coexists with basic perceptual failures",
      "category": "Models & Benchmarks",
      "kind": "Benchmark synthesis",
      "impactMetric": "Around 50.6% top-model analog-clock reading accuracy",
      "finding": "Stanford contrasts extraordinary competition math with brittle performance on ordinary perception.",
      "possible": "AI evaluations should include mundane tests where mistakes matter.",
      "caution": "The abilities of AI remain sharply uneven across task types.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "olympiad reasoning coexists with basic perceptual failures models & benchmarks around 50.6% top-model analog-clock reading accuracy",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-031",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Only a small minority of teachers find AI policies clear",
      "category": "Education & Learning",
      "kind": "School policy survey",
      "impactMetric": "About 6% report clear school AI rules",
      "finding": "Stanford finds substantial ambiguity around what students and teachers can do with AI.",
      "possible": "Clear policies and teacher training are urgent school operations needs.",
      "caution": "Survey findings vary by district and sample.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/education",
      "organization": "Stanford HAI",
      "access": "School policy survey",
      "keywords": "only a small minority of teachers find ai policies clear education & learning about 6% report clear school ai rules",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-002",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Organizational use spreads while fully autonomous agents lag",
      "category": "Economy & Labor",
      "kind": "Organizational survey",
      "impactMetric": "88% of surveyed organizations use AI",
      "finding": "Stanford reports AI use in at least one function becoming commonplace while agent adoption remains single-digit in many functions.",
      "possible": "Organizations can redesign routine tasks but should measure outcomes separately.",
      "caution": "Use does not imply profitability or fully automated processes.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/economy",
      "organization": "Stanford HAI",
      "access": "Organizational survey",
      "keywords": "organizational use spreads while fully autonomous agents lag economy & labor 88% of surveyed organizations use ai",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-047",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Public confidence trails expert enthusiasm about AI",
      "category": "Society & Culture",
      "kind": "Public opinion survey",
      "impactMetric": "73% of experts vs 23% public expect positive work impacts",
      "finding": "Stanford describes a major difference in sentiment about AI and employment.",
      "possible": "Institutions will need transparent evidence about what AI actually changes.",
      "caution": "Opinions are not forecasts and depend on survey wording.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report",
      "organization": "Stanford HAI",
      "access": "Public opinion survey",
      "keywords": "public confidence trails expert enthusiasm about ai society & culture 73% of experts vs 23% public expect positive work impacts",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-034",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Reported AI harms grow quickly",
      "category": "Safety, Trust & Policy",
      "kind": "Incident database analysis",
      "impactMetric": "362 recorded incidents compared with 233 in 2024",
      "finding": "Stanford summarizes reported AI failures and misuse incidents.",
      "possible": "A growing incident corpus can support more systematic response and governance.",
      "caution": "Counts are affected by media attention and reporting rates.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai",
      "organization": "Stanford HAI",
      "access": "Incident database analysis",
      "keywords": "reported ai harms grow quickly safety, trust & policy 362 recorded incidents compared with 233 in 2024",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-023",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Research-replication agents remain unreliable",
      "category": "Science & Discovery",
      "kind": "Benchmark synthesis",
      "impactMetric": "Below 20% on paper-scale replication tests",
      "finding": "Stanford documents low success rates on reproducing research workflows.",
      "possible": "Scientists should demand repeatable experiments, not persuasive AI-written manuscripts.",
      "caution": "A good-looking paper may hide failed research steps.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "research-replication agents remain unreliable science & discovery below 20% on paper-scale replication tests",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-037",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Responsible-AI governance becomes an organizational function",
      "category": "Safety, Trust & Policy",
      "kind": "Organizational survey",
      "impactMetric": "17% more dedicated AI governance roles",
      "finding": "Stanford reports growing adoption of policy frameworks and internal oversight positions.",
      "possible": "AI governance is moving into operations and compliance.",
      "caution": "Having a role does not imply effective enforcement.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai",
      "organization": "Stanford HAI",
      "access": "Organizational survey",
      "keywords": "responsible-ai governance becomes an organizational function safety, trust & policy 17% more dedicated ai governance roles",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-038",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Safety defenses weaken under adversarial prompts",
      "category": "Safety, Trust & Policy",
      "kind": "Red-team evaluation",
      "impactMetric": "Safety benchmark scores degrade under attack",
      "finding": "Stanford reviews jailbreak results and safety limitations.",
      "possible": "Agentic systems need defense-in-depth and access controls outside the model.",
      "caution": "Performance depends on system architecture and attack technique.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai",
      "organization": "Stanford HAI",
      "access": "Red-team evaluation",
      "keywords": "safety defenses weaken under adversarial prompts safety, trust & policy safety benchmark scores degrade under attack",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-024",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Scientific agents trail expert PhDs on complex research tasks",
      "category": "Science & Discovery",
      "kind": "Benchmark synthesis",
      "impactMetric": "38.8% versus 83.5% expert baseline on PaperArena",
      "finding": "Stanford reports a large gap between frontier systems and domain experts.",
      "possible": "AI can assist experts without replacing their verification.",
      "caution": "One benchmark cannot describe every area of science.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "scientific agents trail expert phds on complex research tasks science & discovery 38.8% versus 83.5% expert baseline on paperarena",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-021",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Scientific AI publications surge across disciplines",
      "category": "Science & Discovery",
      "kind": "Bibliometric analysis",
      "impactMetric": "Around 80,150 natural-science AI papers in 2025",
      "finding": "Stanford reports a 26% year-over-year increase in scientific AI publications.",
      "possible": "AI-assisted research is now a broad scientific activity rather than an isolated niche.",
      "caution": "Paper volume does not equal validated discoveries.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/science",
      "organization": "Stanford HAI",
      "access": "Bibliometric analysis",
      "keywords": "scientific ai publications surge across disciplines science & discovery around 80,150 natural-science ai papers in 2025",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-017",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Small specialized biology models beat giant systems on some tasks",
      "category": "Biology & Medicine",
      "kind": "Benchmark synthesis",
      "impactMetric": "111-million parameter model exceeds some larger baselines",
      "finding": "Stanford notes domain-specific models outperforming much larger systems on selected protein tests.",
      "possible": "Small specialist models may be more efficient than a generic frontier model.",
      "caution": "Ranking depends on task, data and experimental setup.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Benchmark synthesis",
      "keywords": "small specialized biology models beat giant systems on some tasks biology & medicine 111-million parameter model exceeds some larger baselines",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-032",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "University computer-science enrollment shifts",
      "category": "Education & Learning",
      "kind": "Enrollment data",
      "impactMetric": "11% drop at US four-year institutions",
      "finding": "Stanford reports lower US CS enrollment alongside growth in specialist AI graduate programs.",
      "possible": "Curricula and career planning should distinguish engineering fundamentals from AI specializations.",
      "caution": "Enrollment change cannot be attributed solely to AI.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/education",
      "organization": "Stanford HAI",
      "access": "Enrollment data",
      "keywords": "university computer-science enrollment shifts education & learning 11% drop at us four-year institutions",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-042",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "US and China competition spans research, patents and models",
      "category": "Infrastructure & Energy",
      "kind": "Research landscape synthesis",
      "impactMetric": "China leads in some patent and publication counts",
      "finding": "Stanford notes growing global scientific competition despite concentrated frontier training.",
      "possible": "Companies should consider multiple ecosystems for research and innovation.",
      "caution": "Quantity is not equivalent to quality or adoption.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report",
      "organization": "Stanford HAI",
      "access": "Research landscape synthesis",
      "keywords": "us and china competition spans research, patents and models infrastructure & energy china leads in some patent and publication counts",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-013",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "US authorizes hundreds of AI-enabled medical devices",
      "category": "Biology & Medicine",
      "kind": "Regulatory evidence",
      "impactMetric": "258 FDA device authorizations",
      "finding": "Stanford reports a growing regulated medical-device market for AI functionality.",
      "possible": "More specialized AI products can be evaluated within existing regulatory pathways.",
      "caution": "Authorization is not proof of improved patient outcomes in every setting.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Regulatory evidence",
      "keywords": "us authorizes hundreds of ai-enabled medical devices biology & medicine 258 fda device authorizations",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Independent evidence",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    },
    {
      "id": "impact-annual-016",
      "date": "2025",
      "precision": "year",
      "year": 2025,
      "month": "2025-annual",
      "title": "Virtual-cell models begin forecasting interventions",
      "category": "Biology & Medicine",
      "kind": "Research synthesis",
      "impactMetric": "Evo 2, STATE and AlphaGenome among examples",
      "finding": "Stanford describes biological models that try to predict molecular and cellular responses.",
      "possible": "Scientists can narrow experimental search before laboratory testing.",
      "caution": "Predictions are not verified treatment effects.",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/medicine",
      "organization": "Stanford HAI",
      "access": "Research synthesis",
      "keywords": "virtual-cell models begin forecasting interventions biology & medicine evo 2, state and alphagenome among examples",
      "reviewed": "2026-11-07",
      "evidenceLevel": "Measured trend",
      "sourceContext": "Year-wide 2025 data as summarized in a 2026 report; no month inferred."
    }
  ],
  "watchlist": [
    {
      "id": "watch-01",
      "horizon": "2027+",
      "title": "AI agents completing reliable multi-app projects",
      "category": "Agents",
      "question": "Watch validated end-to-end task completion, authorization and rollback on realistic workloads.",
      "evidenceNeeded": "Production evaluations and independently documented success/failure rates.",
      "caution": "Not a forecast that reliable full autonomy is achieved in 2027.",
      "status": "Watchlist — not an announced discovery"
    },
    {
      "id": "watch-02",
      "horizon": "2027+",
      "title": "More accessible physical robotics",
      "category": "Robotics",
      "question": "Track generalization beyond staged lab tasks, safe intervention and deployment in real environments.",
      "evidenceNeeded": "Independent field studies and repeatable robotics benchmarks.",
      "caution": "Do not infer commercial readiness from highlight reels.",
      "status": "Watchlist — not an announced discovery"
    },
    {
      "id": "watch-03",
      "horizon": "2027+",
      "title": "Verifiable provenance for generated answers",
      "category": "Infrastructure",
      "question": "Watch whether model identity and source lineage can be cryptographically checked.",
      "evidenceNeeded": "Auditable protocols, deployments and published third-party evaluations.",
      "caution": "Research prototypes may not become widespread products.",
      "status": "Watchlist — not an announced discovery"
    },
    {
      "id": "watch-04",
      "horizon": "2027+",
      "title": "AI-supported scientific experiments",
      "category": "Science",
      "question": "Track closed-loop experimental systems from hypothesis to measured laboratory outcomes.",
      "evidenceNeeded": "Peer-reviewed replications and controls.",
      "caution": "AI-generated hypotheses are not discoveries until confirmed.",
      "status": "Watchlist — not an announced discovery"
    },
    {
      "id": "watch-05",
      "horizon": "2027+",
      "title": "AI media with longer temporal consistency",
      "category": "Creative AI",
      "question": "Track reproducibility, character consistency, audio sync and rights management.",
      "evidenceNeeded": "Benchmarked samples and published access conditions.",
      "caution": "Visual demos may omit failures or editing steps.",
      "status": "Watchlist — not an announced discovery"
    },
    {
      "id": "watch-06",
      "horizon": "2027+",
      "title": "Outcome-based AI automation benchmarks",
      "category": "Business",
      "question": "Track workflows measured by verified time savings, error rates, cost and supervision.",
      "evidenceNeeded": "Independent ROI reports with full cost accounting.",
      "caution": "No guaranteed productivity increase is implied.",
      "status": "Watchlist — not an announced discovery"
    }
  ]
}