Lessons from an artificial intelligence pioneer: Naveen Rao’s guide to the future

Artificial intelligence (AI) is taking over our personal and professional lives at an unanticipated rate. But how do we prepare for an AI future? Naveen Rao, Vice President and General Manager of Artificial Intelligence Products Group (AIPG) at Intel Corporation, tells Vision why – and how –  we should embrace the technology 

Far from its imagined beginnings as a futuristic sci-fi technology, artificial intelligence has become a crucial part of the way we analyse data, run business, and even conduct our personal lives.

Neuroscientist, processor architect and entrepreneur Naveen Rao is an expert in the field – his unprecedented AI business, Nervana, was purchased by Intel in 2016. Now Vice President and General Manager of Artificial Intelligence Products Group (AIPG) at Intel Corporation Intel Corporation, Rao offers expert insight about the technology and its fascinating future.

Naveen Rao is Vice President and General Manager of Artificial Intelligence Products Group (AIPG) at Intel Corporation
Naveen Rao is Vice President and General Manager of Artificial Intelligence Products Group (AIPG) at Intel Corporation

Can you explain for the layperson what the Nervana platform is and what it will enable?

The field of deep learning and artificial intelligence is complicated. It’s got to a point where anyone can take these tools and apply them to data. Part of what we’re trying to accomplish it to make it easier to use these tools and advance those capabilities for companies out there looking to improve their businesses by using AI.

I took an approach that’s not dissimilar from other industries. If you look at web development, for instance, from 20 years ago until today, what has really improved is the tools. The ability to use frameworks, for example, to embed things like rich multimedia experiences [into websites]. Now, it’s standard practice to use these tools for any industry wishing to have a web presence. In much the same way, I see this happening in the AI space. We want to make the tools better, easier to use, so that other companies can have access to them.

How has your work as a neuroscientist influenced your career in deep learning?

Deep learning is ‘Neural Networks 3.0’. Neural networks, as the name suggests, had their starts and underpinnings in neuroscience and our understanding of the brain – they are a mathematical abstraction of how the brain does computation.

I was a computer architect for about 10 years. I worked in Silicon Valley at various startups, and I actually went back to get a PhD in Neuroscience after ten years in industry to understand more about what we know about the brain, and try to apply that back to computers. That’s where we are now – it’s one of those cases where it worked out somewhat according to plan!

In terms of eventual capabilities, how far down the line is AI technology?

We’re just beginning the journey, I would say. Many of the things we called artificial intelligence before really weren’t – they were really just creative kinds of algorithms.

Now, we’ve uncovered a few basic computational principles of the brain that are actually very useful. An analogy I like to use is: birds fly around for thousands of years, and there were many attempts at making a flying machine. But it wasn’t until we understood the principles of flight that we could start building these things. And once we understood those basic principles, it was a pretty rapid evolution. The first aeroplane came out in 1903, and in World War I we were using aeroplanes in warfare, so it was a span of, maybe, 20 years. I think we are actually at that precipice [with AI], where we understand those basic principles. We can go very fast now; it’s going to happen in five to 10 years, whereas coming up with those basic principles takes 20-40 years of research. That’s what I think is very exciting about this time.

What are the key challenges as related to AI?

There are a few. In terms of technical challenges, computational elements – microprocessors – they have to evolve. That’s what I’m trying to do [with Nervana]: change the fundamentals of computation, to make it better for these purposes. I would say that the biggest computation problem we have today is finding useful structure and data.

On the business front, it’s really about finding the business models that work. This is very specific to each area, healthcare being a very challenging one for instance – trying to apply new techniques to radiology or to healthcare records. From a technical perspective it’s actually not terribly difficult – I think it’s a very solvable problem. From the perspective of a business model, it’s actually quite challenging, if you look at where the incentives lie in these industries and things like that.

In other industries like autonomous driving, we have a lot of things to figure out from a legal perspective – where the liabilities lie when there’s an accident, having a mix of human drivers and autonomous drivers… how do we figure this out? There are a lot of challenges ahead, but a lot of it is not technical anymore.

What benefits can AI bring to the professional world?

The benefits that I can identify now are probably just going to be touching the tip of the iceberg. Some of the things are obviously freeing us from the drudgery of repetitive work, just as the industrial revolution freed us from physical labour. What we call knowledge, or thought labour, will shift into more of a creative regime, letting the machines do the more mundane work.

I think that’s the general shift, and obviously that’s going to change what we expect from services and response times, and capabilities of products we buy. I think that it will be a shift well beyond anything we can imagine.

In a hundred years, or fifty years, there will probably be a change in what we call ‘human’ as we start getting better at implanting devices

And will it affect our personal lives?

It already has. We all use Facebook [which changed] the way that we interact with people, and the way that we get our news sources. There was a big kerfuffle in the US elections over ‘fake news’; the way an algorithm on Facebook behaves in terms of screening these things can actually affect the outcome of an election, which can affect global policy! So there’s a perfect example of how this can happen.

Should we fear AI technology?

I think that a lot of this fear is rooted in a lack of understanding of what AI is, and how far along we really are. We’re not at a point where we can build a fully autonomous agent that can go into the world like any other human can. I don’t know that we’ll actually ever build such a thing, because there isn’t a true purpose for doing that. We’re still in control of these things – they’re tools.

I look at an AI agent as something that’s built for a purpose and helps me do a task more quickly and more efficiently. Just like a bulldozer can push a million pounds of dirt, which I can’t do with my own hands, I can now get through a hundred terabytes of data, whereas before I couldn’t do that.

In a hundred years, or fifty years, there will probably be a change in what we call ‘human’ as we start getting better at implanting devices and improving our capabilities directly. At some point in the future we will probably merge with these capabilities, what we call humanity. For the next ten or fifteen years, however, there’s really nothing to worry about.

What can people do to prepare for and adapt to AI both at home and in their professional lives?

Professionally, people can develop a skill. In the past, one person learned from the next in a kind of apprenticeship model, and skills were passed down. The world is going to evolve at a much quicker rate now: if you’re in the technology industry, you have to evolve your skills ever ten years or so. The world is not going to stay the same, and people should get into the habit of daily learning. I don’t think most humans do that very well.

One of the big questions I have is: how is humanity going to deal with the accelerated pace of innovation? I guess that impacts one’s personal life as well.

You were in Dubai recently for the Dubai World Cup. How much do you know about AI in the emirate?

I know that there are a lot of people thinking about healthcare applications in Dubai; that seems to be a big focus that I’ve seen. The way regulations work in Dubai is very different to the US – I think it’s a little bit less constrained – so It could be a really interesting place to try some of these models out. We might be able to figure out some technologies that work and improve people’s health, and if that can be shown in a smaller market like Dubai, perhaps that can be scaled up to the European Union, or China, or something like that.

I met with some of the Dubai 100 teams [a 100-day programme designed to accelerate the growth of early-stage healthcare startups] when I was in Dubai. It was very interesting; a lot of the teams were from places like China and other areas, and there’s definitely a critical mass building.