VIDEO | Novartis Global Head: Start Small, Pick What Works

(US and Canada) Ravi Prasad, Novartis Global Head, Data and AI Strategy and Operations, speaks with Girish Bhat, Acceldata Senior Vice President Marketing, in a video interview about the approach to data architecture, AI/ML application, adoption of the technology and ways to overcome resistance.

Sharing his suggestions for organizations building their data architecture, Prasad advises that they should start simply but agilely. He urges them to understand the organization’s current state of infrastructure or data pipelines and then build the right models around them before signing with a large vendor.

Prasad suggests the “start small, pick what works” approach to demonstrate value and then expand it across the organization. Regarding the technology side, he says that it is a combination of open models and licensed software. Similarly, companies can also consider end-to-end stacks with native cloud providers. 

When asked about Novartis’ investment in ML and its critical lessons, Prasad reveals that the organization is exploring a couple of targeted AI/ML applications. One is a small-molecule lead optimization engine integrated into medicinal chemist workflows. The other application is a knowledge graph-based peer-to-peer video content-sharing platform for physicians.

From a best practices perspective, Prasad describes the deep collaboration with business stakeholders and addressing business needs as critical. Revealing his approach to AI application, he says that he starts with augmented intelligence rather than purely going through an artificial intelligence model.

Explaining the rationale behind this approach, Prasad highlights two things to keep in mind: 

1. Augmented intelligence requires minimal change to adopt. Sometimes data scientists and AI/ML engineers build complicated models to adopt, requiring a significant amount of end-user change.

2. Keep the process extremely agile and pivoting based on feedback from end-users.

According to Prasad, resistance to modern technologies like AI and ML  usually occurs when people do not understand what they are trying and how it impacts their work. He notes that the most significant resistance to AI/ML adoption and data-driven decision making arises from the lack of early engagement from the business.

Prasad suggests two key steps to overcome this resistance:

1. Consistently build communities of practice with business stakeholders and help them understand how AI/ML can help them with their work.

2. Showcase the positive impact on the top line and bottom line.

CDO Magazine thanks Ravi Prasad for sharing his expertise with the broader community.

Read more from Ravi Prasad

Success of Data transformation Lies in Adoption

Data Needs to Be Highly Accessible