At Testlify, we are deeply committed to upholding the highest standards of ethics in artificial intelligence (AI). Recognizing the crucial role AI plays in talent skill assessment, we ensure our technologies are developed and implemented responsibly, with a focus on fairness and transparency.

In line with New York City's Local Law No. 144 of 2021, which requires bias audits for Automated Employment Decision Tools (AEDTs), Testlify pledges to regularly evaluate our AI systems. This law emphasizes the importance of assessing potential disparate impacts on different demographic groups, ensuring our tools do not inadvertently perpetuate biases in hiring or promotion processes.

Testlify’s approach to ethical AI revolves around careful design, continuous monitoring, and proactive governance. We recognize that AI systems can have far-reaching implications on candidates' careers and clients' decision-making processes. Therefore, we commit to a rigorous audit and improvement cycle, ensuring our AI systems are not only efficient but also equitable and just. As of now, Testlify conducts rigorous internal audits of our AI systems. We aspire to elevate our commitment to ethical AI by undertaking an independent audit in 2024.

Testlify’s AI system is at the core of our talent skill assessment platform. It utilizes advanced algorithms to evaluate and score candidate abilities, ensuring businesses find the best fit for their needs. Our system processes a wide array of data to make informed predictions about candidates' skills and potential job performance.

To uphold our commitment to fairness and equity, Testlify’s AI system is designed with several safeguards:

Exclusion of Sensitive Information: We consciously exclude any personally identifiable information (PII), as well as race, gender, and other sensitive data, from our AI algorithms. This ensures that our assessments are based solely on candidates' skills and abilities, not on personal characteristics.

Feature Normalization: Our AI system normalizes features to minimize bias arising from different question difficulties. This ensures that all candidates are evaluated on a level playing field, irrespective of the specific questions they encounter.

Human-in-the-loop Oversight: To avoid over-reliance on automated processes, Testlify incorporates human review into its decision-making framework. While our AI provides valuable insights, final decisions on candidate suitability are made considering a comprehensive human review. This approach allows us to leverage AI’s efficiency while maintaining human judgment and empathy.

The Process Audit framework, as outlined here, represents our planned approach for the independent audit scheduled for 2024. Testlify adopts the Process Audit framework, a rigorous, impartial evaluation methodology to ensure effective management of AI risks. This approach is inspired by financial auditing practices and provides a structured process for AI system audits.

1. Scoping: This initial phase involves a comprehensive survey of Testlify's AI system. The objective is to gain a full understanding of the system's capabilities and limitations, which guides the subsequent audit process.

2. Evaluation & Verification: In this critical phase, Testlify submits documentation that demonstrates our AI system's compliance with established audit criteria. Auditors evaluate and verify this evidence, ensuring it meets the necessary standards. This phase includes:

3. Certification: Upon satisfying the audit criteria, the final phase involves drafting the audit report and certifying the AI system. This certification is a testament to Testlify's commitment to ethical AI practices.

Testlify is committed to ensuring that our bias testing and related procedures strictly comply with established criteria. Our responsibilities in this regard include:

The auditor's primary responsibility is to provide an objective opinion on Testlify's AI system's bias testing. This includes:

To uphold the integrity of the audit process, it is imperative that the auditor remains independent. This independence is in line with standards set by ForHumanity and Sarbanes-Oxley, ensuring that the audit's outcome is unaffected by external influences or conflicts of interest.

It is important to clarify that the findings presented herein are based on internal audits conducted by Testlify. These audits are a part of our ongoing efforts to ensure our AI systems align with our ethical standards and operational goals.

Looking forward, Testlify is committed to conducting an independent audit in 2024. This will provide an external perspective and further validate our AI practices. The findings and opinions from this upcoming independent audit will be shared upon its completion.

Following a comprehensive audit of Testlify’s AI system, the findings are as follows:

The audit concludes that Testlify’s AI system adheres to the required standards for bias and fairness, passing all evaluated criteria. This demonstrates our commitment to ethical AI practices and continuous improvement in our systems.

In the pursuit of unbiased AI, Testlify recognizes the inherent challenges related to the datasets used for bias testing:

While our internal audit results are promising, we acknowledge that the conclusions are subject to these dataset limitations. Testlify remains committed to continually improving our data collection and analysis methods to enhance the accuracy and reliability of our bias testing.

As we conclude this internal audit cycle, Testlify rearms its commitment to ethical AI and responsible technology use. Our successful audit is a milestone, not a final destination, in our journey towards AI excellence.

Looking ahead, Testlify will:

We invite feedback and collaboration from all stakeholders, including clients, candidates, and industry experts, to help us continuously improve and uphold our standards in AI ethics.