AI applications and AI for science

Publishing authors

Base scenario: 35k publishing authors. Paper count × 5.75 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 5,676 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 18,458 papers (5.4%). 8.7% of authorship records lack an Author ID; 15.5% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 5.75; 95% bootstrap interval 5.38–6.14. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

Two related industry segments may overlap. The lower bound is the larger segment; the upper bound is their sum.

Base case: $13.0B.

Pharmaceutical: $10.6B × 100.00%
Biotech: $4.84B × 100.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

KDD 2026 — research topics ICML 2026 — research topics

Scope and related research

Organizations

Selected examples

Google DeepMind Research
AlphaFold predicts protein and biomolecular structures; AlphaFold 3 was developed with Isomorphic Labs.
Isomorphic Labs Product
AI drug design.
Recursion Product
AI drug discovery.

More organizations (8)

Absci Product
AI biologics design.
Boltz Product
Biomolecular models.
Chai Discovery Product
Molecular foundation models.
Generate Biomedicines Product
Generative protein design.
Insilico Medicine Product
Generative drug discovery.
Microsoft Research
BioEmu models protein conformational ensembles.
Nabla Bio Product
Protein design.
Owkin Product
AI for biomedicine.

AI in healthcare

Medical imaging, clinical data, decision support, and personalized treatment.

Health AI Medical imaging

101,106 papers

2.1× vs. 2023

Authors · scenario: 102k–512k
Investment · benchmark: $11.8B

Sources and methodology

Publication activity

2019: 13,081
2020: 22,112
2021: 29,686
2022: 36,563
2023: 48,164
2024: 66,519
2025: 101,106

A broad domain query that includes applied medical research using AI.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("medical" OR "clinical" OR "healthcare" OR "diagnosis") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence") OR ("language model" OR "large language model" OR "vision language model"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 171k publishing authors. Paper count × 5.07 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 5,039 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 101,106 papers (1.0%). 10.1% of authorship records lack an Author ID; 17.6% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 5.07; 95% bootstrap interval 4.73–5.42. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

A published figure for a related Quid industry segment. Its boundaries do not exactly match this research area.

Base case: $11.8B.

Medical and healthcare: $11.8B × 100.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

ICML 2026 — research topics CVPR 2026 — research topics ACL 2026 — research topics

Scope and related research

Organizations

Selected examples

Aidoc Product
Clinical imaging AI.
PathAI Product
AI pathology.
Abridge Product
Clinical documentation.

More organizations (6)

Ambience Healthcare Product
Clinical AI.
Nabla Product
Clinical assistant.
Owkin Product
AI for biomedicine.
Paige Product
AI pathology.
Part of Tempus
Qure.ai Product
Medical imaging AI.
Viz.ai Product
Clinical AI.

AI for physics and engineering

Physics-informed neural networks, neural operators, surrogate models, inverse problems, and physical control.

AI for science Neural operators Physics-informed ML

4,277 papers

2.8× vs. 2023

Authors · scenario: 3.5k–16k
Investment · scenario: $716M–$2.86B

Sources and methodology

Publication activity

2019: 42
2020: 164
2021: 418
2022: 850
2023: 1,535
2024: 2,485
2025: 4,277

Matches include papers applying these methods. Search precision and recall have not been systematically measured.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("physics informed neural network" OR "neural operator" OR "scientific machine learning" OR "physics simulation") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 5.3k publishing authors. Paper count × 3.75 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 3,516 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 4,277 papers (23.4%). 11.2% of authorship records lack an Author ID; 36.2% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 3.75; 95% bootstrap interval 3.57–3.92. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

An assumed allocation of broader investment segments. The range is 0.5–2× the base case. Shares are editorial assumptions, not measured deals or research spending.

Base case: $1.43B.

AI infrastructure/models/research/governance: $143.2B × 1.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

ICML 2026 — research topics KDD 2026 — research topics

Scope and related research

Organizations

Selected examples

PhysicsX Product
AI for engineering physics.
BeyondMath Product
AI engineering simulation.
Luminary Cloud Product
Engineering simulation.

More organizations (2)

Neural Concept Product
Geometric deep-learning models approximate engineering simulations and predict design performance.
NVIDIA Tools & infrastructure
PhysicsNeMo provides tools for physics-based machine learning and simulation.

AI for materials and chemistry

Interatomic potentials, materials discovery, property prediction, and experimental planning.

AI for science Materials discovery

4,116 papers

2.5× vs. 2023

Authors · scenario: 4.3k–19k
Investment · scenario: $716M–$2.86B

Sources and methodology

Publication activity

2019: 466
2020: 717
2021: 927
2022: 1,227
2023: 1,664
2024: 2,403
2025: 4,116

Matches include papers applying these methods. Search precision and recall have not been systematically measured.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("materials discovery" OR "materials design" OR "interatomic potential" OR "materials science" OR "chemical property") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 6.3k publishing authors. Paper count × 4.56 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 4,281 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 4,116 papers (24.3%). 9.4% of authorship records lack an Author ID; 25.0% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 4.56; 95% bootstrap interval 4.34–4.80. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

An assumed allocation of broader investment segments. The range is 0.5–2× the base case. Shares are editorial assumptions, not measured deals or research spending.

Base case: $1.43B.

AI infrastructure/models/research/governance: $143.2B × 1.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

ICML 2026 — research topics KDD 2026 — research topics

Scope and related research

Organizations

Selected examples

Google DeepMind Research
GNoME uses graph networks to identify candidate stable inorganic crystals.
Microsoft Research
MatterGen generates inorganic materials with target property constraints.
Citrine Informatics Product
Materials informatics.

More organizations (4)

CuspAI Product
AI materials discovery.
LG AI Research Models
EXAONE and scientific AI.
Matmerize Product
Polymer informatics.
Orbital Industries Product
AI-assisted materials, hardware and manufacturing development.

Weather, climate and Earth observation

Weather forecasting, climate modeling, remote sensing, and geospatial analysis.

Weather AI Earth observation Geospatial AI

20,029 papers

2.2× vs. 2023

Authors · scenario: 18k–88k
Investment · scenario: $706M–$2.82B

Sources and methodology

Publication activity

2019: 2,401
2020: 3,772
2021: 5,280
2022: 7,042
2023: 9,242
2024: 13,083
2025: 20,029

Covers Earth observation, weather and climate. Incidental mentions may be included.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("weather forecasting" OR "climate" OR "remote sensing" OR "earth observation") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 29k publishing authors. Paper count × 4.41 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 4,315 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 20,029 papers (5.0%). 10.3% of authorship records lack an Author ID; 21.9% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 4.41; 95% bootstrap interval 4.17–4.69. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

An assumed allocation of broader investment segments. The range is 0.5–2× the base case. Shares are editorial assumptions, not measured deals or research spending.

Base case: $1.41B.

AI infrastructure/models/research/governance: $143.2B × 0.50%
Energy management: $4.64B × 15.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

ICML 2026 — research topics CVPR 2026 — research topics

Scope and related research

Organizations

Selected examples

Google DeepMind Research
GenCast research uses generative models for ensemble weather forecasting.
NVIDIA Research
Research on AI models for weather and climate prediction.
Microsoft Models
Aurora is a foundation model for atmospheric prediction.

More organizations (4)

Brightband Product
Earth-system AI and weather forecasting.
Jua Product
Weather models.
Tomorrow.io Product
Weather intelligence.
WindBorne Systems Product
Atmospheric sensing and weather models.

Mathematics and formal verification

Theorem proving, autoformalization, formal specifications, and program verification.

AI for mathematics Theorem proving

834 papers

6.0× vs. 2023

Authors · observed: 3,592
Investment · scenario: $215M–$859M

Sources and methodology

Publication activity

2019: 39
2020: 54
2021: 48
2022: 65
2023: 140
2024: 407
2025: 834

Matches include papers applying these methods. Search precision and recall have not been systematically measured.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("theorem proving" OR "formal proof" OR "autoformalization" OR "mathematical reasoning") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence") OR ("language model" OR "large language model" OR "vision language model"))

Retrieved 2026-09-26.

Publishing authors

3,592 distinct OpenAlex Author IDs across all 834 matching papers. This is an observed count within the query result; no publication-rate assumption is used.

Full query result: 834 of 834 papers (100.0%). 15.4% of authorship records lack an Author ID; 4.3% of retrieved works lack an abstract. Retrieved 2026-09-26.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

An assumed allocation of broader investment segments. The range is 0.5–2× the base case. Shares are editorial assumptions, not measured deals or research spending.

Base case: $430M.

AI infrastructure/models/research/governance: $143.2B × 0.30%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

ACL 2026 — research topics AAAI 2026 — AI taxonomy

Scope and related research

Organizations

Selected examples

Google DeepMind Research
AlphaProof and AlphaGeometry research tackles formal reasoning and mathematical proofs.
Harmonic Models
Mathematical reasoning.
Imandra Tools & infrastructure
Theorem-proving tools check mathematical properties of code and specifications.

More organizations (2)

Axiom Math Models
Mathematical AI.
Microsoft Tools & infrastructure
Z3 supports formal verification of programs and logical properties.

AI in finance and risk

Credit scoring, fraud detection, forecasting, causal effects, and decision optimization.

FinTech Fraud detection Risk modeling

23,621 papers

3.3× vs. 2023

Authors · scenario: 11k–55k
Investment · scenario: $6.52B–$9.11B

Sources and methodology

Publication activity

2019: 2,125
2020: 3,073
2021: 3,923
2022: 4,850
2023: 7,104
2024: 12,410
2025: 23,621

Matches include papers applying these methods. Search precision and recall have not been systematically measured.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("financial" OR "finance" OR "credit scoring" OR "fraud detection" OR "stock prediction") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence") OR ("language model" OR "large language model" OR "vision language model"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 18k publishing authors. Paper count × 2.33 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 2,296 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 23,621 papers (4.2%). 13.9% of authorship records lack an Author ID; 15.4% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 2.33; 95% bootstrap interval 2.18–2.48. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

Two related industry segments may overlap. The lower bound is the larger segment; the upper bound is their sum.

Base case: $7.82B.

Fintech: $6.52B × 100.00%
Accounting/finance: $2.59B × 100.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

ACL 2026 — research topics KDD 2026 — research topics AAAI 2026 — AI taxonomy

Scope and related research

Organizations

Selected examples

AlphaSense Product
Financial and market research.
Feedzai Product
Financial fraud detection.
Kensho Product
Financial AI.

More organizations (3)

Featurespace Product
Fraud detection.
Part of Visa
Sift Product
Fraud decisioning.
Zest AI Product
AI credit underwriting.

Industrial AI, logistics and energy

Quality control, predictive maintenance, industrial control, manufacturing, and resource allocation.

Predictive maintenance Industrial control

32,291 papers

2.5× vs. 2023

Authors · scenario: 22k–109k
Investment · scenario: $4.53B–$18.1B

Sources and methodology

Publication activity

2019: 3,646
2020: 5,625
2021: 7,642
2022: 9,693
2023: 12,697
2024: 18,629
2025: 32,291

Matches include papers applying these methods. Search precision and recall have not been systematically measured.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("predictive maintenance" OR "manufacturing" OR "logistics" OR "industrial" OR "energy management") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence") OR "reinforcement learning")

Retrieved 2026-09-26.

Publishing authors

Base scenario: 36k publishing authors. Paper count × 3.38 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 3,353 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 32,291 papers (3.1%). 11.0% of authorship records lack an Author ID; 22.0% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 3.38; 95% bootstrap interval 3.22–3.54. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

An assumed allocation of broader investment segments. The range is 0.5–2× the base case. Shares are editorial assumptions, not measured deals or research spending.

Base case: $9.06B.

Internet of things: $14.6B × 30.00%
Energy management: $4.64B × 50.00%
Robotics: $7.84B × 30.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

KDD 2026 — research topics AAAI 2026 — AI taxonomy

Scope and related research

Organizations

Selected examples

Intrinsic Product
Robot software platform.
PhysicsX Product
AI for engineering physics.
Cognex Product
Deep-learning tools for industrial image inspection and recognition.

More organizations (13)

Agility Robotics Product
Warehouse humanoids.
ANYbotics Product
Autonomous inspection robots with localization and environment sensing.
Applied Intuition Product
Vehicle intelligence and simulation.
Apptronik Product
Humanoid robotics.
C3 AI Product
Enterprise and industrial AI.
Falkonry Product
Industrial time-series intelligence.
Huawei Models
Industry foundation models.
Nomagic Product
Warehouse robotic manipulation.
Optibus Product
Transport planning optimization.
Samsara Product
AI for physical operations.
Sanctuary AI Product
General-purpose robots.
Secondmind Product
Data-efficient engineering optimization.
Siemens Product
Industrial AI.

AI for cybersecurity

Threat and event analysis, attack detection, triage, and automated investigation.

Security AI Threat detection

11,826 papers

3.0× vs. 2023

Authors · scenario: 6.1k–30k
Investment · benchmark: $8.42B

Sources and methodology

Publication activity

2019: 1,022
2020: 1,555
2021: 2,179
2022: 2,751
2023: 4,005
2024: 6,812
2025: 11,826

Matches include papers applying these methods. Search precision and recall have not been systematically measured.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("cybersecurity" OR "intrusion detection" OR "malware detection" OR "cyber security") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence") OR ("language model" OR "large language model" OR "vision language model"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 10k publishing authors. Paper count × 2.58 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 2,514 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 11,826 papers (8.5%). 11.2% of authorship records lack an Author ID; 12.2% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 2.58; 95% bootstrap interval 2.44–2.71. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

A published figure for a related Quid industry segment. Its boundaries do not exactly match this research area.

Base case: $8.42B.

Cybersecurity, data protection: $8.42B × 100.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.

Scope and related research

AAAI 2026 — AI taxonomy

Organizations

Selected examples

Microsoft Product
Security Copilot assists security investigation and incident response.
Google Cloud Product
Gemini-powered security workflows support alert triage and threat investigation.
CrowdStrike Product
AI cybersecurity.

More organizations (4)

Abnormal AI Product
Email security.
Darktrace Product
AI cybersecurity.
SentinelOne Product
AI security operations.
Vectra AI Product
Threat detection.

Education and social applications

Tutoring systems, interaction analysis, accessibility, and support for working with information.

AI in education AI tutoring Accessibility

40,299 papers

4.0× vs. 2023

Authors · scenario: 23k–115k
Investment · benchmark: $1.44B

Sources and methodology

Publication activity

2019: 1,917
2020: 3,092
2021: 4,019
2022: 5,037
2023: 10,146
2024: 19,934
2025: 40,299

A broad domain query that includes research on AI use and its effects in education.

Title and abstract matches for articles, preprints and reviews; retracted work is excluded. Versions and overlapping areas may be counted more than once.

("education" OR "intelligent tutoring" OR "learning analytics" OR "educational") AND (("machine learning" OR "deep learning" OR "neural network" OR "artificial intelligence") OR ("language model" OR "large language model" OR "vision language model"))

Retrieved 2026-09-26.

Publishing authors

Base scenario: 38k publishing authors. Paper count × 2.86 known authors per paper ÷ 3 papers per author per year. The range assumes 1–5 papers per author; the base and both ends of the range are at least the 2,837 distinct Author IDs already observed in the sample.

The publication rate is an assumption, not a measured rate for this field. The range is a scenario, not a confidence interval.

Random sample: 1,000 of 40,299 papers (2.5%). 12.9% of authorship records lack an Author ID; 8.6% of retrieved works lack an abstract. Retrieved 2026-09-26.

Mean known authors per paper: 2.86; 95% bootstrap interval 2.67–3.09. This describes sampling variability, not search accuracy or uncertainty in the number of researchers.

Applied coauthors are included. Missing or incorrectly linked author records affect these figures. They do not measure research jobs or everyone working in a field.

Private investment

A published figure for a related Quid industry segment. Its boundaries do not exactly match this research area.

Base case: $1.44B.

Ed tech: $1.44B × 100.00%

Company investment, not revenue or research spending. Ranges are scenarios, not confidence intervals.