Workplace investigators currently stand at a technological crossroads. As Artificial Intelligence (AI) integrates into our workflows, the tools available to gather, transcribe, and analyze evidence have evolved at a pace that exceeds the legal and ethical frameworks designed to govern them, introducing risks to the investigative process.
While AI offers the tempting promise of automated transcription and pattern recognition, it can also misrepresent facts and generate entirely fabricated information. As an investigator, I recognize this technological shift is inevitable, but the question remains: does AI truly enhance our capacity, or does it simply introduce a new layer of administrative burden and legal vulnerability?
Efficiency vs. The Obligation of Accuracy
The temptation to use AI to save time can lure an overworked investigator into a dangerous illusion. While the algorithm can transcribe an investigative interview, the investigator remains ethically and professionally obligated to verify every word. In my view, spending hours double-checking text to ensure a witness was not misquoted does not improve efficiency; it simply doubles your workload.
Furthermore, while a machine can transcribe words, it cannot detect sarcasm, manage a difficult witness, or pivot its strategy in real-time. AI is a superb tool for data management, but does it have the cognitive depth required to ensure an investigation is fair and thorough?
The Risk of Fabricated Information
The greatest risk is the tendency for AI to present distorted or made-up information as fact. Most modern AI tools are powered by Large Language Models (LLMs), which are complex algorithms trained on vast amounts of text to predict the most likely next word in a sequence.
It is vital to understand that an LLM does not know facts; it simply understands the mathematical probability of how words should fit together. Additionally, LLMs prioritize sounding helpful and fluent over being accurate. That is, LLMs are entirely capable of hallucinating[1] events, dates, or even legal authorities that do not exist. For investigators, relying on AI is a profound professional liability.
We have already seen this play out in the Canadian legal system in Zhang v. Chen[2]. In 2024, a lawyer used AI for legal research and submitted two non-existent case citations to the court. The AI had completely fabricated these precedents, leading to significant delays and a court order requiring the lawyer to personally pay the opposing party’s legal costs.
In this case, there is direct applicability for workplace investigators. Much like a lawyer, an investigator’s primary value lies in their ability to provide a report that is grounded in evidence. If your report relies on AI-generated logic, you may be adopting fabricated information as your own professional finding. Justice Masuhara highlighted this exact danger in Zhang, noting:
“As this case has unfortunately made clear, generative AI is still no substitute for the professional expertise that the justice system requires… The integrity of the justice system requires no less.”
It cannot be overstated that the integrity of an investigation, and the careers of those involved, cannot be left to a tool that prioritizes fluid sentences over factual accuracy.
The Witness Box Test: Can AI Defend Itself?
Whether an investigator uses AI to passively gather data or actively processes interview notes through the algorithm for expedience’s sake, the evidentiary burden stays the same: the investigator must verify everything.
If your final report relies on AI sorting or automated summaries, it must be able to withstand cross-examination by a third-party examiner or opposing counsel.
In the final analysis, an algorithm cannot sit in a boardroom or step into a witness box to defend its methodology. It cannot explain the nuances of why it flagged one specific phrase as deceptive or why it determined one individual more credible than another. Considering that AI cannot articulate its own reasoning, it cannot defend its own integrity. If you leave your critical thinking to a machine, you leave your entire case open to challenge.
The Rising Value of the Human Investigator
It is my view that the deeper AI integrates into investigative processes, the more valuable human intuition becomes. While technology excels at combing through data, it cannot conduct a trauma-informed interview, read emotional nuances, or balance complex workplace dynamics. That is to say that transforming raw evidence into a just and fair outcome remains a uniquely human skill set.
Ultimately, protecting workplace integrity requires human wisdom, not automated shortcuts.
About the Author

Devan Corrigan is an expert in workplace investigations and labour relations, bringing over two decades of human resources management to his practice. Since founding his independent consulting firm in 2017, Devan has specialized in conducting objective third-party workplace investigations into high-stakes issues, including allegations of harassment, sexual harassment, workplace violence, and complex employee misconduct. Throughout his career, he has been retained to investigate sensitive, high-profile matters carrying significant national implications.
Devan holds a Master of Industrial Relations from Queen’s University, alongside an Honours Degree in Psychology and a certificate in Human Resources Management from Saint Mary’s University. This unique intersection of advanced labour relations expertise and behavioral psychology positions him as a trusted, highly sought-after neutral investigator.
Devan is the lead facilitator for Queen’s IRC’s Fact-Finding and Investigations and Advanced Workplace Investigation Skills programs.
[1] AI hallucinating refers to when an AI system produces information that is false, fabricated, or unsupported by its training data or the given input yet presented as if it were correct or factual.
[2] Zhang v. Chen, 2024 BCSC 285.
Download PDF: Understanding AI in Modern Workplace Investigations


