The secrets of AI success: Using AI to improve employee data accuracy across HR systems

November 10, 2026

Sara Maginn Pacella

Good data leads to better decisions. Effective HR data management is a critical element of data-driven decision-making in human capital management. It supports clear, evidence-based analysis, provides a foundation of accurate data, and contributes to better-informed processes and decisions.

Rebecca Blood, a Chartered Professional in Human Resources, is a human resources consultant at Rebecca Blood Consulting and has over a decade of leadership experience. She spoke with HCM Dialogue to share insights on human resources data quality best practices in the age of AI.

HCM Podcast

Produced with Google Notebook LM Using AI Narration

What are some common HR data quality best practices?

According to Blood, data ownership (outlining who is responsible for what) is key to clear human resources data governance. Blood also emphasizes the importance of conducting regular audits, educating staff on the value of accurate data, consolidating data to reduce duplication and conflicting records, and protecting privacy and security while complying with privacy legislation such as PIPEDA, PIPA and FIPPA. AI transparency is also important. Training should include general information about how AI is used internally and how employee information is collected, used, and shared.

Using AI for HR data management

There are many benefits to using AI human resources software. Using AI for pattern reconstruction and identifying inconsistencies can automate data cleanup, name standardization, classification and other common tasks. Blood notes that a key benefit of AI is cleaner data. She says, “Cleaner data leads to more reliable predictive analytics, which is vital when workforce planning, forecasting, succession planning and conducting compensation analyses.”

While there are many benefits to AI, there are also risks, including the need for validation to prevent bad data from entering the system. Blood also warns that AI can reflect and reproduce historical biases in the organization and its users. Another significant risk that Blood cautions about is hallucinations: instances where AI confidently generates false or fabricated information that is not based in reality. She says, “This is why human oversight and validation remains vital to the successful use of AI.”

” Cleaner data leads to more reliable predictive analytics, which is vital when workforce planning, forecasting, succession planning and conducting compensation analyses ”

Cautionary tales in AI and HR

Blood provides two cautionary tales of AI use in human resources: Workday and Amazon experienced discriminatory biases in their AI-integrated hiring practices. Amazon abandoned an AI recruiting tool trained on a decade’s worth of applicant-submitted data after it learned to favour male candidates because the historical data largely reflected male applicants. Similarly, Workday is facing a lawsuit alleging bias in its AI-assisted applicant screening process. Blood sees this as a call to action for human capital management leaders: “While these examples are in the U.S., Canadian employers should take even more care with their use of AI given that Canada has far more explicit and protective privacy frameworks.”

The importance of human intervention

In late 2025, Gartner reported that “at least 50 per cent of generative AI projects were abandoned after proof of concept due to poor data quality, inadequate risk controls, escalating costs or unclear business value.”

These risks are why businesses are opting for AI systems that operate with human-in-the-loop (HITL) or human-on-the-loop (HOTL) processes. A human-in-the-loop process is when an AI system is designed to stop and wait for human intervention or approval before proceeding with a decision. While the AI operates independently in a human-on-the-loop system, a human is monitoring the process and can override the AI’s decision when necessary.

Medium provides the example of media sites like Facebook and Twitter that use AI for initial filtering of potentially harmful content; however, they still employ human reviewers to ensure that free speech remains protected and harmful content is removed. They also cite a medical AI system that detects breast cancer tumors with a 94.6 per cent accuracy, which increases to 99.5 per cent accuracy with human intervention and monitoring.

These same principles can be applied to human resource systems to improve data accuracy and support informed decision-making.

HR data governance and technology

Blood reiterates the importance of using AI as a decision-support tool rather than a replacement for workplace practices. This includes auditing AI output and monitoring for bias, particularly where this may impact protected groups under human rights legislation. Humans should also review data collection and what is shared with AI tools to maintain compliance with privacy requirements.

She sees the balance between automation and oversight as “just like how we ‘Google’ questions, but we validate the information, or consider its source, and make our own educated and informed decisions based on the information we’ve learned using our own sound judgement.”

What are your thoughts on