Mumbai: Andrew Ng is one of the world’s foremost experts on Artificial Intelligence (AI). He is currently co-chairman and co-founder of the online learning platform, Coursera, and an adjunct professor at Stanford University’s Computer Science Department. He previously served as chief scientist at Baidu and was also the founding lead of the Google Brain team.

In an interview on Monday, on the sidelines of the Nasscom India Leadership Forum (NILF) 2018 in Hyderabad, Ng said life-long learning is becoming the new constant in today’s digital world. He also shared his perspective on the potential of AI and the fears surrounding it. Edited excerpts:

You currently run companies like Coursera, landing.ai, deeplearning.ai and even an AI fund. What’s your goal?

Today, we see Google as a great AI company and Google Brain played a large role in that transformation. Similarly, Baidu is a great AI company in China and the Baidu AI group, which I led, helped in that transformation. But I’m done transforming large companies. To achieve its full potential, AI has to touch companies in other sectors too.

With Coursera, deeplearning.ai and landing.ai, we are trying to help thousands learn AI and, in turn, transform their companies. Landing.ai is helping large companies, especially in the manufacturing sector. The AI fund (raised $175 million till date) focuses on building AI start-ups from scratch. It will be very hard to build a new cloud or mobile company today because of the presence of well-entrenched companies. The space of opportunity, hence, is bigger in AI than in mobile or cloud.

You have often said AI is the new electricity. How would you rate its progress?

We are in the early phases of this (AI-led) transformation. Say, for instance, you have an injection moulding machine. There are a lot of decisions like what are the heating and cooling temperatures, what is the pressure, etc. AI can make those decisions more systematically. In healthcare, we are beginning to see that AI can read the radiology images better than most radiologists. In education, we have a lot of data and companies like Coursera are putting up a lot of content online. There is a lot more work to be done but the (growth) trajectory is very promising.

How would you define a good AI strategy?

Some CEOs have told me: Give us three years to build our IT by which time we will have good amounts of data to work on AI. That is a terrible strategy. Every company has messy data and even the best of AI companies are not fully satisfied with their data. If you have data, it is probably a good idea to get an AI team to have a look at it and give feedback. This can develop into a positive feedback loop for both the IT and AI teams in any company.

Do you think ‘Compact AI’ (on devices) will help further the growth of AI?

Yes. A lot of computing is shifting to the edge—e.g. cellphones, microchip embedded in the light bulb and I think there is a battle for the edge too. Consider the case for speech recognition (Alexa, Google Home, etc.) or self-driving cars, which is driving innovation on edge computing.

When will we get, if ever, to the point of making AI sentient—when machines become as intelligent, or even surpass, human intelligence?

I’d love to get there and I would love that to happen as soon as possible. It is hard to predict whether this will happen in 50 years or even 500 years.

So you clearly don’t subscribe to the fear of AI overpowering humans any time soon?

I think worrying about sentient AI is like worrying about us overpopulating Mars at some time since we haven’t even landed on that planet. I think it’s great that there are a few researchers worrying about sentient AI. We are investing more to address the jobs’ displacement and reskilling problem (due to AI). The issue of sentient AI is distracting many people from addressing these issues.

What about the impact of automation and AI on jobs?

A lot of jobs will be impacted—like those who work in call centres, drivers (with driverless cars) and radiologists. However, the flip side of all the scary reports of potential job losses is that there will be many jobs that are not at risk (from automation and AI). We need to find enough people with skills sets that fit such jobs.

Do you see governments addressing these policy and reskilling issues?

Different countries have various levels of sophistication when addressing these issues. India, specifically, has a fantastic opportunity to leapfrog because countries with a well-entrenched educational system will have a harder time building a new educational system. Hence, India can maybe embrace scalable digitalization faster than other countries. This is a moment in time when the leadership of country will matter.

Silicon Valley and Beijing are the leading hubs of AI, followed by the UK and Canada. I am seeing a lot of excitement in India, going by the number of people who are taking Coursera courses on AI (300,000 of the 1.9 million people overall). Machine learning is the most popular course for people from India. There is a window of time when India can embrace and capture a large fraction of the AI opportunity. But it will not remain open for ever.

In this digital world, why do you still use a whiteboard?

(Laughs). I have been a teacher for long, and think this is an effective way to explain complex AI concepts.

Close