Data Governance and AI

The Guest

A data leader by day, a musician at night, and an optimist at heart. Tiankai Feng is the Data Strategy and Data Governance Lead at Thoughtworks Europe. Tiankai has been in the data game for over 10 years and is passionate about the human aspect of data - working with his clients on improving communication, collaboration, and creativity in data. He is an absolute delight to follow on social media with his numerous and humorous takes on data, musical parodies, and more and he’s here today to talk about data governance in AI.

Data Governance and AI Oversight

Tiankai begins the interview by defining data governance and how it is contained within a data strategy. 

“I define data governance as making data trustworthy, created, processed and used in a functional capacity,” says Tiankai. “Good data governance is about agreeing on definitions as a team. Data strategy attaches to goals, and a part of that is governance. But getting everyone on the same page is the goal.”

He emphasizes that data governance is crucial for enabling trust in data, which drives value for the organization. Tiankai also highlights the importance of AI governance, as AI models need oversight due to their lack of explanability. He points out that public failures with AI, such as chatbots producing inappropriate content, illustrate the need for a structured approach to anticipate and mitigate risks.

AI Marketing and Data Communication 

The conversation then pivots to discussing the capabilities and limitations of artificial intelligence in marketing and communication. While AI can efficiently manage routine tasks, it struggles with creative and ambiguous tasks that require human understanding and creativity. Tiankai shares his experience using AI to generate content, highlighting that the results are less effective than human-generated content. 

“People are realizing that AI models need governance, and require more human oversight,” explains Tiankai. 

“There are so many failures in GenAI. The main reason behind failures is failure to consider scenarios - or prompts - AI doesn’t have the human sense of truth and right and wrong - it only learns data points, but has no moral compass.

We also discuss Tiankai's unique approach to data communication through music, which he uses to make data more approachable and memorable. Check out his “Data Mesh Musical” on YouTube.

Human-Centric Data Governance 

We then switch to talking about the necessity of human-centric data governance and its importance in enabling data use rather than restricting it.

“For me, the human aspect of data governance is underemphasized,” says Tiankai.

“We can have the best tools, but if we don’t have the right people, attitudes, and mindset, nothing is going to happen.”

Tiankai debunked the idea that people are the problem in data governance, identifying processes and decision-making as the real issues that need addressing. Tiankai suggests  adopting an incremental approach to data governance.

“To have people intrinsically motivated by data governance, is rare. You have to frame the value of data governance in personal capacities and ladder it up to business goals - not just leaving it fixed at the end, but bubbling governance into actionable items.”

To find out if Tiankai prefers cheese or chocolate, how he pronounces ‘data,’ and if he would rather play air guitar or air drums, listen to the entire interview at the links provided here. 

The Links

Tiankai on LinkedIn

Tiankai’s YouTube Channel

LISTEN TO THE FULL SHOW -> Stay tuned, stay curious and subscribe to What Gets Measured on Apple Podcasts, Spotify, YouTube or add it as a Favorite on your podcast player of choice.

‍