USA Trends and Developments Contributed by: Christina Hynes Mesco, Chad Ayers, Kristen Prinz and Amit Bindra, The Prinz Law Firm
More and more, employers are incorporating their own customised AI tools into their human resources and legal departments. Although these tools are designed to be suited to the specifics of the organisation, AI is not yet advanced enough to provide comprehensive assistance for any investigation. Even customised tools lack understanding of the nuances of company culture or the team members that may be involved, thereby limiting its usefulness in drafting a detailed investigation plan or other supports. AI can, however, provide certain outputs, such as an overall map for an investigation, as long as they are reviewed and modi - fied by a human investigator. Employers must be especially cautious about priva - cy when employing AI in investigations. Workplace investigations typically implicate personal and sensi - tive information, and investigators must be aware of the restrictions (or lack thereof) tied to their use of AI tools. Some AI systems are open source, which are generally more vulnerable to malicious actors, as their code is publicly available. Yet even closed-source AI, the source code and algorithms of which are private and proprietary to the developer, present risks and do not alleviate all concerns around data privacy. Even tailor-made AI systems that are internal to an organi - sation are not risk-free. Investigators should gener - ally refrain from inputting privileged or highly sensitive information into any AI system and should also be aware that generative AI outputs may be discoverable in litigation. Humans bring a host of biases to their work, even to supposedly neutral investigative work. Perhaps the best way to incorporate AI into the investigative pro - cess is to use it as a check on one’s own potential blind spots. Reciprocally, a human investigator must always be aware that AI outputs are not simply bereft of error or biases themselves. An investigator must be in an ongoing conversation with AI outputs, mean - ing the investigator constantly and critically assesses both machine- and person-generated content.
AI should be used minimally in defining the issues of an investigation, as bias or misinformation at this point could impact the entire investigation. AI may, however, help with forming questions and identifying witnesses, potentially supplying a broader perspective and help - ing to identify gaps in an investigator’s approach. As always, human discernment should be applied. AI tools can also be used to schedule interviews and to take notes during interviews. Employers need to be aware that jurisdictions have differing regulations related to recordings and many jurisdictions are in the process of passing AI regulations. Even with AI help, investigators should continue to maintain their own notes because AI tools are known for inaccuracies in transcribing information. Further, AI tools can pro - vide summaries of relevant documents and interview notes, especially when these are voluminous. If using AI-generated notes, have someone review them for accuracy before they are relied upon for further steps. Investigators are ultimately responsible for a thor - ough and meticulous investigation, and their final reports should accurately reflect what was discovered through document review and interviews. AI is a great tool, but it is not a replacement for a human touch – at least not yet.
444 CHAMBERS.COM
Powered by FlippingBook