Talking AI with Nvidia, and how deep learning can change the world

During Nvidia’s recent GTCx event in Mumbai we met with Jay Puri, Nvidia’s executive vice president of worldwide field operations, who is responsible for global sales and regional marketing of the company’s products and services. Discussing aspects as varied as the implications of the booming field of AI to the sectors that stand to most benefit, here’s what we spoke of:
 

Which are the sectors that AI will have the most profound impact on?

AI has been around for 50 years -- it has gone into and out of vogue with high expectations that didn’t quite deliver. But most it then had been around rule-based modeling. It’s only recently that the deep learning version of AI came to play: rather than it being rule-based it is now about writing a neural network and feeding it a lot of data where the network can recognize elements better than a human being does. The increased adoption of AI is a combination of a few things: 1) Because of the Internet, the amount of data available right now is much more than ever before, and 2) The GPU with its parallel processing gives the right amount of compute power required to derive insight from this data within a reasonable timeframe. Many of the problems that computing wasn’t able to solve before, which were to do with pattern recognition rather than computing based computing algorithms, is now applicable to just about everything. It still stunning when a car [the BB8] can drive itself based on just experience -- just like how human beings learn.

We’re move into a new era of computing -- the AI era -- and this is going to transform just about every facet of our lives and every industry. Frankly I think it’s going to be completely pervasive. The biggest success we’ve seen are from Google, Facebook, Baidu etc, essentially those that provide Web services and have access to huge amounts of data. Going forward, we’re starting to see it being used in places like healthcare, where it can have the most dramatic impact. Even retail -- from customer insights to inventory management, to finance -- in fraud detection, anywhere that you can recognize patterns and get insight with a lot of data, AI can improve on.
 

How do smaller companies that do not have access to large amounts of data get onto the AI wave?

Because we’re in the early phases of this computing paradigm, the larger companies find it easier to deploy. We’re trying to make it easy for people doing work in this area with a supercomputer that is like an appliance, ready to go out of the box -- the DGX-1. It used to take people a lot of time to put the whole system together, get all the software, load the frameworks, doing data science. Our solution is also available in the cloud in Microsoft Azure, AWS, IBM and so on.

But if you’re a small company and don’t have access to the data for each of the verticals, there are companies starting to develop solutions. Companies like SAP and Salesforce will actually train [neural] networks based on their data and offer those services that other companies, especially smaller ones, can use. Once the industry gets more mature, you won’t necessarily need to train everything on your own data, you’ll be able to get solutions who are building solutions for vertical markets.
 

What do present-day established companies need to do to leverage the offerings of AI?

It’s going to be the same as it has with every new generation of technology, except here it is potentially more massive. For example self-driving cars and the impact on the transportation industry as a whole. Car companies are panicking at this point, especially since Tesla has announced self-driving capability in all of their cars starting now, with full level 5 autonomous driving within 18 to 24 months. Conventional wisdom with most companies put this at the year 2020-21. Now they’re all calling to say they’ve got to speed up their deployment of self-driving cars. The impact on this industry is going to be huge -- unless car companies adapt and find their new place, some of them will go out of business. It’s one of the largest industries that will be disrupted.

There are many positives from an economic perspective, but there’s the other side -- what happens to the drivers that the cab and truck companies stop employing? I don’t think we have fully understood the full impact because of this AI revolution. The more optimistic view is, whenever there’s been a disruption society has changed to create different type of jobs. Social media, for example, with new jobs on marketing and analysis. Human beings find new ways to be creative, to take on whatever foundation technology provides and find ways to be productive.