VIDEO | Morgan Stanley AI Product Manager: Synthetic Data Protects Data Privacy in Regulated Sectors

(US and Canada) Supreet Kaur, Assistant VP and AI Product Manager at Morgan Stanley, speaks with Deborah Lorenzen, Head of Enterprise Data Governance and Master Data Management, in an engaging video interview, about her role, synthetic data and its relevant business use-cases, ways to unlock its potential, the data bias challenge, and how synthetic data can bring value to the organization.

Kaur joined Morgan Stanley in February 2022. Coming from a data scientist role, she pivoted into a data strategy role, recognizing the importance of data in building AI products. Her role now involves building strategies around existing data to improve AI products.

When asked about synthetic data, she explains that it is created artificially to simulate reality and is derived from existing data. Kaur discusses its usage in data augmentation and how it involves creating rather than finding data in a database.

Referring to use cases, Kaur notes that synthetic data protects data privacy in regulated sectors such as healthcare and finance. She also suggests that organizations use synthetic data to make data sets more balanced or to simulate edge cases yet to be seen in historical data. Autonomous vehicles and EVs have been leveraging synthetic data for years to allow for realistic testing, says Kaur.

Delving further, she asserts that three points are necessary for successfully leveraging cutting-edge technologies:

1. Specialized talent - to build and manipulate complex algorithms
2. High-quality data - to mirror real data in synthetic data
3. Data accessibility - to build models

Highlighting the aspect of data bias, Kaur urges data practitioners to ask themselves questions related to the synthetic data they are using. She emphasizes that there are metrics to measure biases in synthetic data; however, they might not be suitable for every use case. She further stresses the need for practitioners to have domain knowledge other than technical expertise to ask the right questions to create an efficient model.

Moving forward, she discusses AI and machine learning as fields of education and training that require specialized expertise. She observes that it can be challenging to convince senior leadership about using them as there are no 'apples-to-apples’ comparisons for explanation.

To try and tie it to business sense, Kaur suggests using tools such as explainable AI and frameworks to assess the return on investment (ROI) the model can bring. Additionally, she encourages senior leadership to take a genuine interest in learning technical details, as data scientists put effort into understanding the domain. She states that they can upskill if both parties are willing to learn.

Emphasizing the organizational value of synthetic data, Kaur believes it to be a versatile solution for companies ranging from startups to more mature organizations. She mentions that synthetic data can be used for data augmentation, which is especially useful for FinTech startups that have limited customer data but want to build AI-driven solutions.

Additionally, synthetic data can also come to the rescue for organizations attempting to detect cancer within images and for those edge cases that require advanced specifications. In conclusion, Kaur remarks that any organization must invest in good quality data and skilled talent to manage it before attempting to build AI-driven solutions.

CDO Magazine appreciates Supreet Kaur for sharing data insights and success stories with our global community.

The opinions conveyed in the interview solely belong to Kaur and do not reflect those of her employer.