Reflection from Queens University IRC Webinar
Sundays are a good day to share what I am learning, reading, and reflecting on. For this particular Sunday, I am choosing to share my reflection from Queen’s University IRC Webinar titled Navigating AI in Workplace Investigations, presented by Devin Corrigan and Deborah Hudson.
This webinar was quite timely and needed, especially with the continuous rise of artificial intelligence tools and systems in professional work, including workplace investigations. I am particularly curious about AI not simply from the perspective of what the technology can do, but also what its growing use means for human rights, privacy, procedural fairness, professional responsibility, and governance.
As AI tools and systems become more sophisticated and increasingly embedded in the workplace, investigators will inevitably have to consider where these tools can appropriately be used to support our work and, equally importantly, where they should not.
I will take a pause here to explain the term “Workplace Investigations” for those who may be unfamiliar with the process.
Workplace investigations are formal fact-finding processes used by organizations to investigate complaints or concerns about conduct in the workplace, which may involve allegations of harassment, discrimination, workplace violence, policy breaches, or other forms of misconduct. Investigators are appointed to conduct impartial and comprehensive formal investigations, which may involve gathering documentary evidence, interviewing the parties involved and witnesses, assessing the evidence, and making findings of fact on a balance of probabilities (whether it is more likely than not that something happened) based on the information available.
One of my takeaways from the webinar is that AI can be used to support some parts of the investigation process, such as administrative tasks, but it cannot replace the investigator or the place of professional judgment. The speakers shared during the webinar that generative AI tools and systems can omit important information, misunderstand context, produce inaccurate information, or even generate information that does not exist. This reinforces the need for thoroughness and the use of professional judgment to ensure that investigators go back to the source and verify the information being used.
Another important point raised during the webinar was the distinction between delegating a task and delegating responsibility. There may be parts of the investigative process where technology can reduce administrative work, but there are other aspects that remain fundamentally dependent on professional judgment and human expertise. This is not to say that administrative tasks completed with the assistance of AI should not also be reviewed. Rather, it calls for greater scrutiny, especially when determining what evidence needs to be collected, assessing relevance, evaluating conflicting evidence, making credibility assessments, and ultimately reaching findings of fact.
This is particularly important because formal investigations involve people and may somtimes have significant consequences for everyone involved. The investigation findings may affect someone’s employment status, professional reputation, workplace relationships, and, in some circumstances, lead to subsequent third party proceeding such as legal processes, union griveance. Therefor, an investigator remains responsible for understanding the evidence and being able to explain how they reached their conclusions.
One last point I would like to highlight from the webinar is the importance of maintaining privacy and confidentiality and considering applicable legal obligations when conducting investigations. Workplace investigation files often contain sensitive information including medical information, witness statements, personal information and confidential organizational information. The convenience of uploading an interview transcript or a collection of documents into an AI system therefore has to be balanced against existing privacy, confidentiality and legal obligations.
Investigators need to understand which tools their organizations permit, where information is stored, retained, who may have access to it, and whether information entered into a system may be used for other purposes. These considerations should form part of responsible professional practice when determining whether and how AI can be used.
In closing this post, one area that I want to explore further is how we can navigate the use of artificial intelligence during workplace investigations through a human rights lens, recognizing that we cannot take the “HUMAN” away from the process, including important issues around power dynamics, culture, identity, intersectionality, and other equity considerations. This consideration is particularly important as AI systems are trained on existing information and data that may sometimes reflect biases and inequalities that already exist in society.Applying a human rights lens requires us to ask not only whether these tools and systems are convenient and efficient, but also whether they are sufficiently inclusive and designed to be used in ways that mitigate stereotypes, inequalities, and biases.
In summary, artificial intelligence tools may assist with some aspects of an investigator’s work, but they cannot replace the professional judgment required to recognize the nuances, thoroughness, and specialized expertise needed to produce a comprehensive and detailed final report.
Disclaimer: The views expressed here are my own and are shared for educational and informational purposes only.
CHIDINMA UMAHI NWANKWO 
Leave a comment