Forlais Research
Argus 2
Argus 2 sustained the successor programme, exposed an inherited false positive and carried a verified correction back through the Argus record.
The question
A successor that corrected its own lineage
Argus 2 asked what a research successor should do when a celebrated finding from its own lineage no longer survived scrutiny.
Research journey
How the work unfolded
Argus 2 carried the Argus lineage into a sustained successor run. It preserved recoverable progress across the assessed period and continued through detected performance degradation.
A failed research gate revealed that an inherited signal could appear more convincing as assessment scale grew even when the underlying evidence did not.
The project explained the false positive and corrected it throughout every retained assessment scale. Independent rechecking then matched the corrected outcome.
The correction was carried back into the earlier Argus record. The original operational success remained visible, while the scientific interpretation of its early findings changed.
Turning point
What changed the direction
The failed gate exposed a growing source of false confidence. Argus 2 used that failure to repair the result and the lineage record instead of protecting an attractive conclusion.
Achievement
What Forlais established
Argus 2 established sustained bounded operation for the active successor and a verified correction that held throughout the retained review, creating a safer evidential foundation for the Argus lineage.
Significance
Why it matters for AI
Argus 2 shows why credible AI research must be able to correct its own history as capability grows, with operational success and scientific validity recorded separately.
Public scope
Forlais Research is the public research surface for Forlais Group. Related official routes include the company site, EvaEsi overview, Project Eve, and selected research programme pages.