---
title: "PICARD (Probing Intelligent Capabilities via Artificial Randomized Data)"
type: "entity"
entity_kind: "concept"
status: "watching"
canonical_name: "PICARD (Probing Intelligent Capabilities via Artificial Randomized Data)"
aliases: ["PICARD"]
first_seen: "2026-10-22T00:00:00.000Z"
connects_to: ["RIKER","JV Roig","Kamiwaza AI","contamination-resistant evaluation"]
writer_model: "claude-sonnet-5"
---


JV Roig's 2025 technical-report framework ("Testing what models can do,
not what they've seen," Kamiwaza AI) that RIKER's own paper credits as
providing "the underlying contamination-resistant evaluation
methodology" its "paradigm inversion" approach extends. Not yet
independently read by the vault — flagged as a watching stub from
[[claim-riker-2026-no-independent-source-same-author-admits-no-replication]]'s
capture so the first-seen date is stamped before it is lost; promote to a
full hub if a future session reads the 2025 report directly or finds
PICARD recurring beyond the RIKER lineage.

## References

- [[claim-riker-2026-no-independent-source-same-author-admits-no-replication]]