While A.I development is accelerating at an exciting rate, it’s still not entirely clear how quickly the government sector is picking up on new developments. To learn more about how the government can and will take advantage of machine learning and artificial intelligence, we spoke with Anthony Robbins, federal vice president at NVIDIA.

ABERMAN: Tell us about the government sector. There’s been a lot of talk around I.T. modernization, but it’s not as well known what’s going on in the commercial sector and entrepreneurial community. Unpack that for us. What’s going on exactly?

ROBBINS: There’s a lot going on in artificial intelligence and across the federal government. And of course, as we all know, artificial intelligence is one of the most transformative technology transformations to come into the federal government. In February of this year, the federal government signed into law the executive order on artificial intelligence, and while not actually a law, has the power of law. So as a result, there’s been a bunch of focus on use cases across the federal government on how they can bring artificial intelligence in to serve and provide improvements to citizen services. Just in September, the CTO of the federal government announced a billion dollars to go into research and development.

You know, a year before that, DARPA, through A.I. Next, announced two billion dollars in funding for A.I. Next, which was a program that’s going to fund, 85 different initiatives across five years for DARPA to bring A.I. to the Department of Defense. The DOD CIO has announced it as a priority, and on and on. And so, the point is, artificial intelligence has significant momentum across the federal government. And there’s really great opportunities for us to drive down the overall costs of government, and improve citizen services by use of artificial intelligence.

Subscribe to the What’s Working in Washington podcast on iTunes.

ABERMAN: So, let’s still delineate that a bit, because I think that people who don’t spend a lot of time around A.I. will spend a lot of time confusing what I would call the development of general intelligence, the idea of thinking machines, whether conscious or not, vis-a-vis the specific intelligence, the ability to do a repetitive task of greater complexity better or faster than a human being. My impression is that, what we have now is, within the government, we have both kinds of activity going on. I think about DARPA, I think more about specialized intelligence, a high level or general intelligence. And when I think about government and citizen service, I tend to think about specific intelligence and doing things more effectively. Is that a good way for me for me to mind map this?

ROBBINS: I think so. And let me just try to take what you said and apply it to some use cases. So, a really easy way for the federal government, and for us in industry, to connect to the federal government, is on something called robotic process automation. And that’s just simply taking repetitive tasks that machines might be able to do faster and better than humans. And so, that’s a robotic process automation, RPA. It’s a really important use case. The cost to get into that is not very expensive, and then it allows civilian agencies to get their feet wet with A.I. Then there’s some other use cases, like in the healthcare business. It’s going to be completely transformed by by bringing artificial intelligence into health care. And we’ll all benefit from that.

Radiology, for example, that progress is already occurring there. There’s another area in the case of waste, fraud and abuse. So in the case of waste, fraud and abuse, the federal government spent 141 billion dollars in 2017. And I think the reported number was about 27 billion in the IRS alone. And in that use case, they have vast amounts of data that machines can work through, and find patterns in, doing anomaly detection at a rate that’s much faster than humans. And so, really important to A.I. is vast amounts of data. And the federal government, as the largest producer and consumer of data in the world, probably only second to the cloud service providers, has this richness in data, but they need help in labeling that data, so that these machines could work on that data to add value into a bunch of use cases, some of which I just mentioned.

ABERMAN: So NVIDIA, my awareness of it, I’ll admit, is really great graphics cards. and gaming machines and so forth. How did the company migrate into what sounds much more like a software driven business?

ROBBINS: Yeah, that’s a great question. Twenty five years ago, when we were founded, I don’t think we were founded on the vision that we were going to add value to the world’s progress in artificial intelligence. But what we did do, in the beginning, is we built world class capabilities around graphics and video games. And the parallel processing that relates to processing vast amounts of 3D graphics and video, as it turns out, is similar to what is needed to process vast amounts of data to train these neural networks, which is the fundamental aspect of artificial intelligence. So our GPU is kind of at the heart of the world’s progress in artificial intelligence.

We came out with that 20 years ago for graphics and video games, and then several years after that, we came out with a programming model that allowed, as mere mortals, to access the hardware. So once we gave engineers and data scientists around the world the ability to access the GPU, or graphics processing unit, we found breakthroughs beginning to occur in the area of artificial intelligence. And literally, that window was around 2012. It’s acknowledged in the industry. In 2012, there was some breakthrough work in computer vision that literally accelerated the world’s progress on artificial intelligence. And here we are, seven years later, making a ton of progress.

ABERMAN: It is interesting to me how that story bears out something that I feel really strongly about, which is that: all the focus we have in our society on entrepreneurial behavior in the startups loses that most people, like yourself and your colleagues, are actually entrepreneurial within an existing organization, intrapreneurs, building new products. You’ve had a long career working in government, selling into government, working education. What’s your view of innovation sourced by large companies? What do they have to do well in order to be innovative?

ROBBINS: Well, that’s interesting. So, if I answer the question from the perspective of NVIDIA, Jensen Huang is our CEO, and he’s brilliant beyond description. And the way that he manages our company, it’s as if we’re a bunch of startups inside of NVIDIA. And NVIDIA is, you know, 13,000 employees, 13 billion in revenue. But the way we operate, and so I have responsibility for the public sector, and as we look at this public sector marketplace, we act like a startup inside of this big company. So I think that our company is agile, we’re nimble, we’re flexible, we move fast. We think a lot about innovation. We challenge the customer. We challenge our federal government.

In the case of artificial intelligence, the government is moving faster than they have in other technology adoptions. We think they need to be moving faster. So, I think because of how our company operates, we do a good job acting like a startup. And I think we add value, then, to the federal government’s mission. We also work with startups. So, we have an inception program, where we work with 5000 startups who are doing work in and around artificial intelligence, and accelerated computing, and graphics and the like. So, we try to support the startup community or the innovators. We act like an innovator ourselves. And I think therefore, as we face our customer, we give them that innovative spirit.

ABERMAN: I think that brings me to something that’s been scratching my head. I hear that 70, 80 percent of the government I.T. spend is on maintaining legacy systems. You know, I remember a couple of years ago I read a story about how there were still eight inch floppy disk drives in minuteman silos, which terrified me in a very deep way. But how is it possible? The Obama administration and Trump administration, are they creating models that make it easier for government to innovate, or is it really still outside the private sector pulling folks in innovation?

ROBBINS: Well, I think it’s some of both. And in the case of artificial intelligence, the ability for us to be good service to the federal government requires us to play the greatest team sport we’ve ever played in participating in the transformation in the federal government. Because higher education and universities have to play a role, the innovation community, startups have to play a role. Defense contractors and large systems integrators have to play a role, as well as the federal government. The federal government cannot outsource the work that they need to do in A.I., because they have to participate. The transformation itself is so big, they have to have subject matter expertise in this area. So civilian agencies, Department of Defense, the intelligence community, systems integrators, start ups, defense contractors, universities all have to come together to figure out how we can help our government accelerate, so we can go address this A.I. for good, and improvement of citizen services.

And so, it’s a giant team sport now. I think the government’s done a pretty good job here. In the case of the Department of Defense, they released an A.I. strategy. The Air Force has released an A.I. strategy. SOCOM’s released, or is working on, an A.I. strategy. The intelligence community has published a document. Their aim, augmenting intelligence with machines, publicly available. The executive order, as I mentioned earlier. So, I think the government is doing a pretty good amount of work projecting to industry some of the needs that they have. And I think it’s incumbent upon us to make sure that we do a good job listening, and then crafting strategy that we can support this marketplace.

ABERMAN: Anthony, your excitement is palpable, which is always fun to have a guest who’s excited about what they do. What’s the coolest thing you’ve seen over the last year or two in A.I., using your company’s technology or elsewhere, where you thought, oh my God, I’m living in the future?

ROBBINS: There’s this statement that often gets made that, if we have kids and grandkids, you know, that they actually will never have a driver’s license. And so, the work that’s going on in autonomous systems, the transportation industry as a whole is a 10 trillion dollar market opportunity, that will be transformed by artificial intelligence. As a result, there’s a lot of early work going on in autonomy and autonomous systems. So, people look at that and they say, okay, we’re gonna have autonomous things like helicopters and cars, and our kids may not have a driver’s license. Or if they do, it’ll be just to put it on the wall and make fun of mom and dad, kind of like what we used to have in phone booths and the like.

So, the other thing that’s really interesting is: this A.I. transformation that’s underway, in the case of the federal government, it’s almost not about the technology any longer. Although, it is about change and transformation. And in many cases, that’s about people adopting something new. And in the case of artificial intelligence, people have concerns about ethics and security, different use cases, which, you know, which we’re all aware of. The question, as we face the federal government, we talk to them as much about how they lead change and transformation as it relates to artificial intelligence. Because the technology has proven itself in use cases in most countries, and companies all around the world, at a rate that was probably not forecasted.

ABERMAN: Well, thanks for coming in today and sharing some of your enthusiasm from a government perspective around A.I. Anthony Robbins, VP of North American Public Sector at NVIDIA. Thanks for joining us.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.