Findings
We publicly test how AI systems behave in Polish and publish results together with the method and the data.
Latest Laboratory finding
Not published yet
The first Laboratory finding will appear here
We have not published any findings yet and we do not show results we do not have. When the first one is ready it will open this page: with the thesis, the method, downloadable data, limitations and the history of corrections.
Method, no secrets
How we work
We will publish every finding in the same order, in five parts:
- Thesis: one sentence: what we found and who it matters to.
- Method: how we measured, in enough detail that it can be repeated.
- Downloadable data: all responses and test suites in open formats.
- Limitations: what this finding does not prove.
- Corrections: every change after publication, with date and reason.
This order is a design assumption of the portal, not a description of existing results.
Transparency and independence
The Foundation is independent of public authorities, political parties and businesses, in particular providers of AI systems that may be the subject of its work. We publicly disclose our sources of funding, organisational structure and statutory tasks — details on the Finances page.
The portal does not use cookies and does not track visitors — see the Privacy Policy.
Three directions
What we will do
Laboratory
measurements of AI systems in Polish: test suites, scorecards, open data.
Planned for later stages.
Observatory
AI in public administration: freedom-of-information requests and a register of systems.
Planned for later stages.
Help desk
support for people harmed by a decision made by an AI system.
Planned for later stages.
Everything the foundation has published is in the publications catalogue.
Questions about findings or method: Contact.