Join John Thompson, the Head of AI at EY and an expert in Generative and Causal AI, as he discusses the intricate balance between privacy and productivity in the workplace. He dives into the challenges of implementing generative AI in a regulated environment while ensuring data security. John also shares insights on enhancing efficiency through AI in tasks like document summarization and explores the potential of synthetic data. His seasoned perspective on industry trends and decision-making in AI is both enlightening and essential for any tech enthusiast.
Read more
AI Summary
Highlights
AI Chapters
Episode notes
auto_awesome
Podcast summary created with Snipd AI
Quick takeaways
Implementing generative AI at EY necessitates balancing privacy and productivity by creating a secure environment and limiting logging features.
Generative AI is transforming workflows by enabling faster access to information, though true productivity metrics are still challenging to quantify.
Deep dives
Adopting Generative AI in a Regulated Environment
Implementing generative AI within a highly regulated organization like EY requires a strategic approach to address privacy and security concerns. John Thompson, the head of AI at EY, highlighted the challenges faced when integrating AI technologies into the workflow of 420,000 employees. The focus was on ensuring that sensitive data could not be compromised while allowing employees to benefit from productivity gains. This involved creating a secure Azure environment and disabling logging features that could expose client information, illustrating the balance between risk management and innovation.
Exploring Employee Use Cases for Generative AI
EY employees are leveraging generative AI across diverse applications, from summarizing complex financial reports to generating proposals. The technology facilitates quick access to vast amounts of information that were previously underutilized, enabling employees to work more efficiently and creatively. Despite the potential for productivity improvements, true quantitative assessments remain a challenge, as many are still in experimental phases and rely on subjective self-reports. This illustrates that while early adopters enjoy some benefits, widespread productivity gains are yet to be fully realized.
The Debate on Data Logging and Its Implications
The decision not to log interactions with generative AI tools at EY has sparked discussions about accountability and data utility. While this approach mitigates privacy risks, it limits the ability to analyze user interactions for performance insights. Consequently, a debate is ongoing about the merits and risks of implementing logging systems to facilitate data-driven decisions while balancing the organizational need for privacy. This situation underscores the complexity of navigating data governance in a rapidly evolving technological landscape.
Future of Work and AI in Decision-Making
The conversation around AI's influence on the future of work suggests a shift towards more fulfilling roles as repetitive tasks become automated. Generative AI can assist in simple decision-making processes, while complex decisions may benefit more from causal AI applications, which seek to identify underlying factors affecting outcomes. As AI systems evolve, particularly with advancements in agent technology, organizations can optimize processes like travel planning by automating decisions based on set parameters. This transition indicates an exciting evolution in job functions, emphasizing that enhanced intelligence will create opportunities for higher-level creative work.
By now, many of us are convinced that generative AI chatbots like ChatGPT are useful at work. However, many executives are rightfully worried about the risks from having business and customer conversations recorded by AI chatbot platforms. Some privacy and security-conscious organizations are going so far as to block these AI platforms completely. For organizations such as EY, a company that derives value from its intellectual property, leaders need to strike a balance between privacy and productivity.
John Thompson runs the department for the ideation, design, development, implementation, & use of innovative Generative AI, Traditional AI, & Causal AI solutions, across all of EY's service lines, operating functions, geographies, & for EY's clients. His team has built the world's largest, secure, private LLM-based chat environment. John also runs the Marketing Sciences consultancy, advising clients on monetization strategies for data. He is the author of four books on data, including "Data for All' and "Causal Artificial Intelligence". Previously, he was the Global Head of AI at CSL Behring, an Adjunct Professor at Lake Forest Graduate School of Management, and an Executive Partner at Gartner.
In the episode, Richie and John explore the adoption of GenAI at EY, data privacy and security, GenAI use cases and productivity improvements, GenAI for decision making, causal AI and synthetic data, industry trends and predictions and much more.