A practical look at governing AI in threat assessment work. When two models...

https://romeo-wiki.win/index.php/What_Are_the_Risks_of_False_Positives_in_Workplace_Threat_Assessment%3F

A practical look at governing AI in threat assessment work. When two models disagree on a risk rating, the disagreement gets logged instead of averaged away. Each assumption behind a finding is written down so a reviewer can challenge it later