Top 5 Lessons Learned from CDAOs Successfully Implementing GenAI

As reported by Gartner in 2023, out of the 70% of organizations exploring generative AI, only 19% had piloted or productionalized models. Although 68% of those executives believed that the benefits outweigh the risks, Gartner Group Chief of Research and Distinguished Analyst Frances Karamouzis cautioned, “Initial enthusiasm for a new technology can give way to more rigorous analysis of risks and implementation challenges. Organizations will likely encounter a host of trust, risk, security, privacy, and ethical questions as they start to develop and deploy generative AI.”

Most if not all of the CDAOs who successfully piloted or productionalized GenAI did encounter those questions, and eight have responded to the authors with interesting insights.

Without good data, there is no AI…

...said Heidi Lanford, former CDO of Fitch Group and Red Hat. “It’s pretty simple. You cannot have good AI without a solid data foundation.” Walgreens CDAO Vipin Gopal noted that “many organizations are rapidly trying to adopt GenAI but your data environment must be ready to enable value creation from AI.” An Oil and Gas CDO noted, “AI doesn’t make magic out of it. It’s only as good as the data you put into it.”

Conclusively, success favors organizations with maturity in data aggregation and democratization, data management, data governance, and above all, data quality. Data quality is essential because GenAI models create and make decisions. Poor data quality not only affects model output which engenders mistrust, but it also confounds the ability to understand how the output was derived.

GenAI is risky business

Lanford held a mirror to model mistrust to demonstrate its converse — the misuse of model output by decision-makers. “If you use natural language generation and you’ve never written anything, how can you critique or judge or approve a screenplay that AI generated? You’ve never written, you don’t understand what good writing looks like or how it’s done.”

Similar to Karamouzis, Gopal indicated “the vital need to address the core risks while tapping into the potential of the technology: ethical, data privacy, regulatory, and security.” Education is a key risk mitigation for Gopal.

Education is essential

“There’s understanding the technology and training the technical folks, but more important than both is the interpretation of what the model is doing,” said Lanford. “If business stakeholders aren’t aware of what to look for regarding bias and quality, then GenAI models could be misused and misinterpreted, and the business units acting on the insights won’t achieve the desired result, leading them down the wrong path.”

Gopal recommends educating at all levels. In his prior role, he launched the Eli Lilly Data and Analytics Institute, a data literacy program to educate over 40,000 employees on how data and analytics technology could improve their day-to-day responsibilities. “It’s no longer about teaching technical people about data and AI, but teaching the entire organization, especially business leaders who will be making decisions based on AI models and AI output.” The curriculum will vary based on the job responsibilities.

Choose to buy or build tools carefully

For organizations whose prompts can be answered with internet data and who have no fear of exposing their interactions externally, low-cost solutions like ChatGPT are appropriate. A McKinsey poll revealed that Marketing, Sales, and Product/Service units regularly use such tools to gain efficiency and that the top use cases are document drafting, trend analysis, and chatbots.

Most enterprises pose questions that the Internet cannot answer; furthermore, they are unwilling to expose sensitive customer and financial data externally. Some have chosen to build their own LLMs with costs ranging between US$ 2M-3M. Others have opted to plug open-source models and tools into their secure data platforms such that models can be trained safely.

Weighing risks and rewards can be challenging and may necessitate a new breed of AI leaders.

Leadership and accountability are essential

AI must be responsibly led, but in addition to data and technical acumen, AI leaders must be business-savvy to ensure the proper usage of AI insights by decision-makers. Board Member Elena Alikhachkina asked, “Who is explaining the future? A business strategist should lead AI, someone with a clear picture of the future company: how they will operate, make money, and land customers. Chief Strategy Officers should become AI leaders.”

Another successful GenAI implementer believes that CDAOs should not only lead the technical components but also bring to the role a sense of what is real or not real, what is achievable, and be able to plot value creation on an AI roadmap – much bigger remits than supplying technical AI expertise.

CDAOs who have successfully deployed GenAI have established the proper leadership accountability to mitigate security and privacy risks through education and thoughtful tool selection. Unsurprisingly, data is key — high-quality enterprise data that is kept safe and secure. 

About the authors:

Shayde Christian is Chief Data and Analytics Officer at Cloudera. He guides data-driven cultural change for Cloudera to generate maximum value from data. He enables Cloudera customers to get the absolute best from their Cloudera products such that they can generate high-value use cases for competitive advantage. Previously a principal consultant, Christian formulated data strategy for Fortune 500 clients and designed, constructed, or turned around failing enterprise information management organizations. Shayde enjoys laughter and is often the cause of it.

Richard Pooley is a Partner in Odgers Berndtson’s Boston office and a member of the Global Technology Practice. He focuses on recruiting Chief Data Officer, Head of AI, Chief Technology Officer, and Chief Information Security Officer roles for a broad range of organizations, ranging from privately-owned/PE-backed businesses to publicly owned global corporations across multiple industry sectors.

Prior to joining Odgers Berndtson, Pooley was a member of the Global Technology and Professional Services Practice at Heidrick & Struggles, and before this, he was a member of CTPartners’ Professional Services and CFO Practices.