Cyvaa Research

Empirical AI Evaluation Across Civilizational Texts

Systematic documentation of citation drift and hallucinated commentary across philosophical corpora.

What you'll find here

Empirical studies, shorter notes, and the methods behind them — all published openly

Research

Field Notes

Methodology

Empirical studies of how language models cite texts and attribute scholarly positions. Full methods and data published with each one.

Shorter pieces — a single documented model failure examined closely, or notes on other researchers' work in this area.

How the tests are designed and scored. Methodology is published before results, so findings can't be selected after the fact.

Our Approach

Our focus remains strictly empirical: tracking citation drift in complex philosophical corpora.

Cyvaa Research

Receive Research Dispatches

Stay informed with our latest findings on AI and civilizational knowledge, delivered directly to your inbox.