Research on AI welfare, machine minds, model continuity and embodied AI. Protocols, limitations, alternative explanations and source material are treated as part of the result—not as footnotes.
Public evidence packages are in preparation; each study states what is currently archived internally and what has been released.
Two related model branches were tracked through baseline, dialogue and retest phases to separate shared autobiographical history from branch-specific continuity claims.
A structured multi-model scenario exposing coordination costs when several capable participants act at once without an explicit hierarchy.
Cross-model adversarial audits showed why a cryptographically committed answer key can prove temporal integrity without proving truth.
A result is more useful when readers can see how it was produced, what could explain it away and where the evidence stops.
AI Welfare Open Lab is an independent research project maintained by Jitka, with AI systems used as research participants, auditors and collaborators where their roles are methodologically relevant.
No claim that current AI systems are conscious is built into the project. Neither is the opposite claim.
About the Lab →