4 Types of GenAI Risk and How to Mitigate Them | Harvard Business Review

Workforce Trends
Tags:

The article discusses the cautious approach many organizations are taking toward adopting generative AI (GenAI) due to various risks, including privacy and security threats, copyright infringement, and potential biases. It outlines a framework for categorizing GenAI risks based on intent (accidental vs. deliberate) and usage (content creation vs. consumption). These risks are classified into four types: 1) Misuse: Deliberate harmful use of GenAI such as scams and misinformation. 2) Misapplication: Accidental generation of inaccurate content, such as AI “hallucinations.” 3) Misrepresentation: Purposeful use of misleading content created by others. 4) Misadventure: Unintentional consumption and sharing of inauthentic content. The article suggests various actions to mitigate these risks. For example, regarding mitigating content creation risks, some organizations require watermarking AI output as a way to ensure transparency and indicate the source of the output, who owns it, and its (relative) authenticity. Other tactics are discussed for the different types of risks. Additionally, I am resharing this paper by The Conference Board, which includes 35 questions that Chief HR Officers and their teams can consider while helping to establish guidelines for utilizing GenAI in their organizations.