(US & Canada) VIDEO | Good AI Needs Right Data — McKesson Head of Decision Intelligence for US Pharma

Yazhene Krishnaraj, McKesson Head of Decision Intelligence for US Pharmaceutical, speaks with Susan Wilson, Group VP Americas Solution Leader at Informatica, in a video interview about data governance as an enabler, strategies to approach data governance, creating good AI with data, data democratization, and the organizational journey around deploying AI.

At the onset, Krishnaraj declares that data is everywhere and people have access to it, regardless of attempts to restrict it. Therefore, the strategy is to create data governance policies, stating the standards and guardrails. She adds that if businesses are empowered with the right capabilities and guardrails, they will follow and enforce them.

Next, she shares that data governance has become an essential topic, and the organization is bringing in the data governance program carefully, keeping data democratization in mind.

Further, Krishnaraj acknowledges that changing the conventional understanding of data governance is a challenge. Nevertheless, the organizational approach at McKesson focuses on trying to identify the core problems that data governance can alleviate. Krishnaraj notes that from data quality to compliance and even ethical use, businesses need help in multiple areas.

To further illustrate the benefits of governance programs, the organization takes the ‘use case approach’ wherein strategic use cases are created to resolve the most pressing business pain points, allowing businesses to see data governance as a growth enabler.

Moving forward, Krishnaraj states that to create good AI, the right data is necessary. She further states that it is a natural progression to converge data ownership with analytics and AI. As such, it is essential to guarantee that data is not being misinterpreted and that the data owner understands the data and how to interpret it properly.

Furthermore, she encourages data democratization across the business, stating that data owners may know more about data assets, but various business segments may want to use them for different use cases. Krishnaraj concurs that the data owner should not own both analytics and AI but must be data literate, know good analytics, and ensure that value tops the data.

Thereafter, she affirms that McKesson is in the middle of its journey to deploying AI with successful use cases. Reminiscing about the lessons learned, Krishnaraj states that productionizing AI has been the biggest challenge, given the ever-changing nature of business and the difficulty in finding the right talent.

In continuation, she notes that while building models, data scientists are stuck maintaining and tweaking the same model due to a lack of productionization. Therefore, she urges organizations to prioritize productionizing AI so that data scientists can continue building new models.

Additionally, Krishnaraj asserts that the reason behind creating an AI model should be communicated well across the business, as it relates to the value aspect. Moreover, the change management aspect, which includes knowing how the business will change after its deployment, should be monitored, she adds.

Furthermore, Krishnaraj suggests considering its impact on business. She recommends analyzing the efficiency a business gains and the additional insights that are driving revenue. She concludes by stating that data science is not just something that data scientists do; there is a culture and change management part to it as well.

CDO Magazine appreciates Yazhene Krishnaraj for sharing her insights with our global community.