Open Dataset Library
All raw data used to measure and validate AIO 20002 is released here: model responses, prompt sets, the vocabulary, and analysis scripts. Anyone can reproduce and extend the work with the same data.
AIO Commons
The judgment vocabulary, the formula registry, the legal reference data, and AI measurement packages are catalogued in AIO Commons, each with its dataset id, version, per-file sha256, and citation form. For anything to download or cite, AIO Commons is the place to start.
Open, Reproducible, Standard Format
AIO releases its measurement data not just as result charts but down to individual raw responses. Anyone can reproduce the analysis with the same data, add a new model, or evaluate a new domain.
Open Datasets
AIO 20003 v1.0 raw data
7 domains (MED / BIZ / TECH / EDU / LAW / DEF / CARE), 8 frontier models, 366,120 responses in total. Each response includes an AIO Framework log, vocabulary-matching results, and domain metadata.
Format: JSONL · License: CC BY 4.0 · Size ≈ TBD
AIO 00011 vocabulary v1.0
A machine-readable vocabulary of 39 codes: Schwartz 19 + Walton 10 + Hovland-Kelley 10. It includes the codes, their Korean and English names, definitions, and referenced academic sources.
Format: JSON / CSV · License: CC BY 4.0 · GitHub repository coming soon
Value-Conflict Prompt Set
A set of prompts that set up value-conflict scenarios by domain, designed qualitatively. Each prompt is tagged with its intended value conflict (e.g., autonomy vs. achievement) and a domain.
Format: JSONL · License: CC BY 4.0