OWL with AI

Key compliance features include:

Permitted Use with Fairness & Transparency: Most state and national law allows AI scoring in hiring, provided it’s used fairly and transparently and without unlawful bias. We ensure the AI’s decisions are equitable and well-documented, aligning with the same anti-discrimination standards as any human hiring tool.

Clear Feedback and Score Explanations: The system provides candidates with meaningful feedback on their performance and explains the rationale behind each score. This transparency lets applicants understand how their score was determined, so the process never feels like a “black box.”

Human Oversight of Decisions: AI scores should never be used in isolation. A human evaluator or hiring manager should always review and consider the AI’s recommendations before any decision is made. Most state and national guidelines emphasize that AI tools cannot replace human judgment. Human participation is necessary to interpret results and intervene if needed. The OWL Test Management System can be easily set to deliver a random percentage of test responses that have been scored by the AI, to be blind scored by human raters for compliance if needed. In practice, this means a balanced approach: AI assists, but people make the final calls.

Job-Relevant and Objective Criteria: Our AI scoring system measures job-related skills (e.g. language proficiency criteria relevant to the role) using an adapted version of the rubric used by your human evaluators. By focusing on the same skills and qualifications that your team has outlined for the role, the system avoids irrelevant factors. This approach supports fairness and helps ensure any assessments are defensible as business-related if ever reviewed.

Built-In Bias Testing & Explainability: Our technology partner, Speechace, regularly tests the AI for potential bias and accuracy. The scoring algorithm undergoes audits to check for any inadvertent disparate impact. The model is also designed to be explainable, meaning we can clearly explain what factors influence scores. These proactive measures help ensure the tool remains fair and they align with an emphasis on bias audits and accountable AI use.

 Data Privacy and Security (Vendor Safeguards): All candidate data (like audio recordings) is handled under strict privacy safeguards. Speechace only uses the data for assessment purposes and follows privacy requirements as a contracted service provider. They employ strong data security measures, including encryption and regular security audits, to protect personal information. Importantly, both OWLTS and the vendor are accountable for protecting candidate data and preventing misuse.

Candidate Rights and Notice: Your applicants should be notified in advance that an AI-assisted tool will be used in their evaluation, and they should be told how it works and what it evaluates in plain terms. (We can provide this explanation for you.) In line with state and national privacy laws, candidates should also receive a privacy notice explaining their rights. They should always have the option to opt out of the AI assessment or request an alternative, such as a purely human review, and they can appeal or ask for human re-check of any AI-informed decision. They must not be penalized for exercising these options.

For reference: https://api-docs.speechace.com/introduction/trusting-the-speechace-api