Director of AI at Mastercard Addresses The Gap Between AI in Research and AI in Industry

How can we work together to close the gap?


Wow AI

2 years ago | 2 min read

Regarded as‘a disconnection from the industry’, AI research has been reported to not yet evolve into a successful formula for innovation for smaller firms though it has helped big tech firms such as Google, IBM, and Microsoft.  

Is AI research truly difficult to commercialize? How huge is the gap between AI in research and AI in industry and how can we bridge it? 

These questions were briefly discussed by Karamjit Singh during a pre-event interview with Wow AI and will be answered in detail at Worldwide AI Webinar 2022.

Keep reading to learn more about Karamjit and his insights into the matter. 

Don’t forget to watch the whole interview here.

About the speaker

Karamjit Singh is currently working as Director of AI and managing a unit called AI Garage at Mastercard. After taking a Master’s in Computer Application, he spent the first phase of his career focusing on researching artificial intelligence and machine learning. Now going into the second phase, he has shifted his attention to building products using technology.

Previously, he worked in Accenture, UnitedHealth Group, and Tata Consultancy Services. Karamjit has authored over 15 publications in top-tier AI/ML conferences and has over 20 patents filed by his name. 

At the upcoming Worldwide AI Webinar, Karamjit will talk about how industry and research can come together to bridge the gap between AI in Research and AI in Industry.

On the gap between AI in research and applied AI

When asked about the current AI adoption, Karamjit shared that AI has been adopted or at least recognized by every industry and every company wants to build or invest in AI. Since businesses have already realized the immense potential AI can bring about, it’s essential that an accurate AI implementation strategy is used. 

Mr. Singh believed that each company needs to invest in the right infrastructure because building good AI models needs huge datasets and the current infrastructure might not be adequate to facilitate it. He also added that both the company and the developers must clearly understand what needs to be done and make sure they know which steps to be taken to deploy AI. 

While building the model is the easiest part when you have a team of people with great skills and expertise, according to Karamjit, the challenge is to make sure that there is a seamless pipeline from when the model is deployed to when the actual product is out in the market.

“Many industries are definitely investing a lot in AI. But what is lacking is the knowledge to take the model and actually deploy it. You would see that there is a significant difference between the number of models being built versus the number of models being deployed.”

On the future innovations of AI/ML in the banking sector and FinTech

Apparently, Mastercard is leveraging deep learning technologies to solve challenges in the transaction world. 

Karamjit stated that banks have been using traditional machine learning models for transactional data for quite some time without the penetration of deep learning as it is more suited for image and text data. The upcoming graph neural networks that are being implemented into the payment industry will be exploring the relationships between the transactions and definitely give us solutions that were never seen before.

Along with other industries, payments will also evolve and a future where people can use their Mastercards to make transactions through a virtual platform like Metaverse is not far away.

Join our upcoming Worldwide AI Webinarto hear more from Karamjit Singh.

Watch the whole interview here.


Created by

Wow AI

Wow AI is a trusted partner that provides high-quality data for AI training in different languages at a large scale. We are experienced professionals with a track record of success in numerous types of data collection. With a crowd of over 100,000 project contributors all over the world, we can provide massive, scalable, high-quality data across data types. We also provide off-the-shelf data and tailor-made data for your AI/ML projects.







Related Articles