Driving Productivity and Improved Outcomes With Generative AI-powered Assistants 14 Delivering GenAI That Adheres to Principles of Trust, Security, and Responsible Use It is critical to make employees, partners, and customers comfortable that GenAI tools are being used in a trustful, secure, and responsible manner. It is not just a matter of compliance, legal, or governance, but it goes to the very heart of an organization’s brand: Can I trust doing business with you and that you are using my data in a responsible manner? This is especially important as employees “experiment” with GenAI usage on their own, given how intuitive these tools often are and how cost-efÏcient they are to acquire and use. It is important for organizations to take necessary steps to avoid “rogue/shadow AI” or AI sprawl, which entails discovering and mitigating LLM usage to ensure conformity with best practices, corporate governance, and other usage requirements. AI assistants need to be designed and deployed with these principles in mind, including strategies to ensure fair and ethical use of machine learning models. Research from Enterprise Strategy Group noted that 54% of organizations feel it is important to establish and follow ethical guidelines, while 51% indicated they seek to incorporate explainability and transparency.¹¹ Naturally, challenges abound: Most organizations (51%) said they struggle with balancing competing goals of accuracy, performance, fairness, and ethics; 42% said they lack diverse and representative data; and 40% reportedly lack clear guidelines.¹² Challenges Organizations Face When Ensuring Fair and Ethical Use of Machine Learning Models 51% 42% 40% 35% 31% 1e+9d= 2d+8e= 0d+0f= 5c+5f= 1c+9f= Balancing competing goals Lack of diverse and Lack of clear guidelines DifÏculty assessing fairness Lack of standards of accuracy, performance, representative data fairness, and ethics © 2024 TechTarget, Inc. All Rights Reserved. 11Source: Enterprise Strategy Group Research Brief, Navigating the AI Ethics Landscape: The Case for Governance, January 2024. Back to contents 12Ibid.