FairSight

FairSight is an AI Governance and Risk Assessment Copilot built with IBM watsonx Orchestrate. It evaluates any AI project idea and instantly generates a Governance Assessment Report.” “Maya describes her proposed HR promotion model in a single sentence… For example: ‘Analyze an AI system that recommends employees for promotion using past performance reviews, feedback scores, attendance, and manager comments.’” FairSight processes this through a multi-step workflow that extracts key attributes, identifies risks, checks ethical concerns and bias, evaluates data protection, and produces a concise, actionable governance summary.” “FairSight identifies that this system relies heavily on historical performance reviews, manager opinions, and subjective feedback — which means there is a high risk of bias. Existing disparities in performance evaluations could easily be amplified by the AI model.” “It also highlights moderate privacy risk, because sensitive employee data is being processed. Even if the system is used internally, the data must be handled carefully and securely.”