The source index

Every claim should leave a trail.

Here is the trail: every primary source cited on AI Solved X, the databases we scout, and the rules that keep a promising headline from becoming a misleading one.

01.

Discovery is not evidence.

Press releases, news coverage, newsletters, and social posts can point us toward a story. They cannot carry the claim by themselves.

02.

Primary beats popular.

We link the paper, registry, public dataset, technical report, or official evaluation that readers can inspect directly.

03.

Limits travel with the result.

Study design, sample, setting, conflicts, missing data, and real-world distance stay attached to every conclusion.

Where we look first

Ground truth before buzz.

This is a working map, not a claim that every record inside these systems is correct. Each candidate still has to survive our evidence score and independent fact check.

Health and clinical evidence

We start with the published paper, then check trial registration and outcomes when human health claims are involved.

PubMed ↗Biomedical papers and publication records
ClinicalTrials.gov ↗Trial registrations, protocols, and reported results
WHO ICTRP ↗Clinical-trial records from registries around the world
CDC Data ↗Public-health surveillance and official datasets

Research records and reproducibility

These sources help us find early work, verify citations, and locate the code or data behind a claim. A preprint is a lead, not proof.

arXiv ↗Early research signals before peer review
Crossref ↗DOIs, publication metadata, and citation verification
Zenodo ↗Citable research datasets, software, and supporting files
OSF ↗Study materials, preregistrations, data, and analysis plans

Climate, Earth, and infrastructure

For forecasts, hazards, water, weather, and environmental claims, we look for observations and operational data—not a screenshot of a model demo.

NOAA NCEI ↗Weather, climate, ocean, coastal, and geophysical archives
NASA Earthdata ↗Satellite and Earth-observation datasets
USGS Data ↗Water, geology, hazards, ecosystems, and mapping data
Data.gov ↗U.S. government datasets across public services

Energy, transport, and engineering

We favor public test data, technical reports, standards work, and field results over product announcements.

DOE OSTI ↗Energy, computing, physics, and engineering research results
NHTSA Data ↗Vehicle safety tests, recalls, investigations, and crash data
NIST Data ↗Measurement science, standards, benchmarks, and reference data
NASA Science Data ↗Open scientific data across NASA research programs

Currently cited

The live research ledger.

This section grows automatically with the directory. One published brief, one inspectable primary source.

Health · Strong evidence

AI Solved Breast Cancer Screening.

Interval cancer, sensitivity, and specificity comparing AI-supported mammography screening with standard double reading without AI in the MASAI studyThe Lancet · 2026DOI 10.1016/S0140-6736(25)02464-X

Every claim should leave a trail.

Discovery is not evidence.

Primary beats popular.

Limits travel with the result.

Ground truth before buzz.

Health and clinical evidence

Research records and reproducibility

Climate, Earth, and infrastructure

Energy, transport, and engineering

The live research ledger.

Real images, credited.

AI Solved Diabetic Eye Screening.

AI Solved the First Draft of a Heart Scan.

AI Solved Hearing One Voice in a Crowd.

AI Solved Finding Lost Nazca Lines.

AI Solved Hearing Which Bird Is Singing.

AI Solved the Building Blocks of Whale Talk.

AI Solved Better 10-Day Weather Forecasts.

AI Solved the New-Hire Learning Curve.

AI Solved Cutting Flight Contrails.

AI Solved 214 Million Protein Shapes.

Get the brief, not the hype.