AI & ML interests

Open, freely adoptable vocabulary and measurement standards for AI integrity — in the spirit of Unicode and W3C rather than certification bodies. We develop numbered specifications (AIO 20001/20002/20004, 00014) covering shared terminology for model behavior, declaration formats, and reproducible measurement protocols. Work in progress; specifications and evaluation datasets are published openly for review and adoption.

Recent Activity

makarismov  updated a dataset about 2 months ago
aioq/aio-benchmark-judgment-distributions
makarismov  published a dataset about 2 months ago
aioq/aio-benchmark-judgment-distributions
View all activity