Hello and welcome to the NVIDIA AI Podcast.
I'm your host, Noah Kravitz. Most episodes of the AI podcast focus on how individuals and commercial or research organizations are using machine learning, deep learning, and other technologies that fall under the artificial intelligence umbrella.
Today, we're going to widen our scope and examine how countries are adopting national AI plans.
Our guest is Keith Stryer. Keith is Vice President of Worldwide AI Initiatives at NVIDIA, and he's here to talk about how the trend of national AI plans started what these plans can accomplish, and how they should differ from one country to another.
Keith, thanks for taking the time to join the AI podcast.
Oh, it's a pleasure to be here. Thanks. So can you start by telling the listeners a little bit about your work at NVIDIA and also about what the AI Nations Initiative is?
NVIDIA has a long tradition of engaging with the developer community and more recently data scientists as well, regardless of industry.
And we have invested heavily in, software programs and other kinds of outreach efforts to help these technicians do their life's work on our platforms.
AI Nations is one of the growth initiatives that I oversee at NVIDIA.
It's a global initiative to engage with more, we'll call non-technical stakeholders, government leaders. to help to really understand their policy goals and the national plan for AI. and to help them understand accelerated computing and how it can be applied to the individual missions of the ministries within their country, the ministries of transportation and health, and more broadly to drive economic growth, national security, and other key policy priorities.
The NVIDIA narrative is compelling. It's very exciting, but it's also very complex.
I mean, consider The NVIDIA's AI platform operates inside pretty much all the world's largest hyperscale data centers all the major cloud providers.
It's inside the world's fastest supercomputers and scientists in pretty much every field from astrophysics to biotech depend on on the platform to advance their discoveries.
But at the same time, that same platform, NVIDIA's technology also powers racing drones.
Right. You know, and you know, the drone racing league and, and robots to deliver pizza.
I mean, literally, you know, for dominoes and. and autonomous trucks.
I mean, it's the brain inside machines that automate dangerous tasks on oil rigs. helps monitor post-op cardiac patients, detect financial fraud, even measure beach erosion.
So part of my role. is to engage with government leaders and policymakers to really help them understand this incredible narrative and all the ways in which AI and NVIDIA can benefit their country and map to their goals.
The second part, though, is more tactical.
It's really to identify through these engagements, through that conversation, opportunities to forge a special relationship, a formal partnership in many cases with governments They want to harness all of our capabilities to help them with AI education and up-skilling, joint research, boost the potential for their local startup ecosystem, accelerate things like their national genomics program, even the the development of autonomous public transportation systems.
So when we're afforded this opportunity, to play a small role in a country's AI journey.
We respect the impact this will have. It's a purposeful effort. you know, that's really aligned to our mission to solve the world's most complex problems.
And we don't take it lightly. And as you were talking, I was listening and wondering what a sample day on your daily calendar must look like.
It's really boring. Yeah, exactly. So when you use the term national AI strategy, what exactly does that mean?
Well, first, it's a tool for communications in a sense.
It sends a message, just having a strategy.
It sends a message to citizens, to business leaders within that economy, and really to the world.
And the message is this. We understand the importance of artificial intelligence to the future of our nation, and we have a plan.
And that's important, Frank. The process itself can also be just as important as the plan, which is the importance of engaging with stakeholders, the public consultation the transparency that this creates.
I mean, it's more than just a policy writing exercise, right?
Much of the benefit comes from just the collaboration that comes through that.
The process usually starts with a set of basic questions.
How can we as a country leverage AI to advance our national priorities?
And those do differ, right, from country to country.
How do they support our dominant industries?
Certainly, that differs country to country.
And how can it sustain our values? How can we leverage AI, not just as a technology, but as something that can protect?
So countries are going through this exercise in different ways, but the output is roughly the same.
It's a framework. So this strategy is a framework that sets forth priorities. identify those key stakeholders, hopefully identify sources of funding, a timeline, some kinds of targets.
They tend to be more qualitative. They can be quantitative as well.
You might, for example, say, We're going to train 1000 new, you know, PhDs in the fields, you know, in certain fields, that kind of thing.
Economic growth in general tends to be the number one priority for these plans, although there are some that have different focuses.
Some plans put a greater emphasis on national security or health care.
More recently, countries like Norway and Denmark and others have explicitly stated that promoting responsible AI is going to be the defining characteristic of their national AI strategies.
That's more of a recent focus. But really, if you go back three years ago, So this is pretty recent history.
You just go back to June of 2017, there wasn't a country on earth that had this.
Oh, wow. Okay. It's a new chapter. Starting with Canada, Canada was the first.
And now if you jump ahead, there's more than 50 countries that have taken this step. there's still a bunch that are in that sort of discovery phase but haven't published their strategy yet.
But, and some of the early countries like the United Arab Emirates, the UAE, They've actually continued to develop their thinking, which is a good thing, and have actually published sort of an updated strategy, which is a good practice.
Right. Technology in general is obviously a huge part of how Many nations, many people run on a day-to-day basis.
With so many new tech initiatives around the world right now gaining attention, 5G comes to mind, space exploration technology, green tech, clean tech, and so on.
Why are national AI strategies a particular priority?
Yeah, I think that's a great question. And I really do think it's different.
If you look at the last... a few hundred years and up through sort of the industrialization of our economies.
I mean, obviously the biggest countries in the world you know, defined the power base, right?
The geopolitical base and where opportunities came.
And that's just not going to be the same.
It plays out quite differently in this case.
Even smaller countries like Lithuania, Malta, Vietnam.
Serbia, Armenia, just go down the list. Smaller countries, they have all developed national AI strategies.
They're all making investments in national AI infrastructure.
And they will actually... have their place on that map.
And so I think it is very different. And that's because the impact of AI can be seen across the economy. in ways that range from being very cute, frankly, having toasters you can talk to and little robots to entertain you, up to mind-blowing use cases.
There's plenty of hype too. I mean, the real challenge here is more about misunderstanding, really attributing capabilities to AI that just aren't realistic and not appreciating the potential even within the limits.
But AI is changing the nature of work itself, right? not just one industry, but all of work through the automation of work in a way using human-like cognitive functions that weren't possible at scale until very recently.
And it's changing fields like entertainment through real-time ray tracing and the ability to create these hyper-realistic imagery and scenes without any time for rendering. manufacturing, agriculture.
I mean, you just go down the list, right?
So I think between 2016, 2017, again, this is recent history.
It was almost as if there was a collective realization by the governments worldwide that that AI is more than just an academic discipline.
It's more than a technology. It's more than just a service within the tech sector.
AI is an opportunity that needs public leadership.
And I think it needs public leadership for two reasons.
One, AI is changing the basis for competition on a global scale.
I mean, if you do not have domestic AI infrastructure and domestic AI capabilities and a domestic AI ecosystem, I believe the countries will be less relevant.
They will have more difficulty keeping entrepreneurs at home, attracting researchers, supporting long-term economic growth.
So there's a strong imperative in that sense.
The other is that some AI applications do raise some complex social and ethical questions.
They introduce new risks that need to be studied, managed, and in some cases, even regulated.
So this is an important role for government, balancing... that need for innovation and to avail ourselves of this 21st century technology, but also while addressing humanity's biggest problems, being mindful of some of the downsides.
And that's I think that the nature of it has really empowered countries essentially all at once, again, with a very short period of time. almost always at the head of state level to take this on.
So from global superpowers, right, the big countries, to small island nations, really everyone's on board at this point.
Hearing you talk about how, on a national level, nations will need infrastructure that supports AI to compete going forward?
Maybe think of the term digital divide and just how much more meaning a term like that takes on and can be applied to different situations as we move forward through time.
It's really something to think about. Absolutely.
Kind of along those lines, you talked about how AI affects all of these different industries and walks of life.
And certainly we've explored that on this podcast and obviously NVIDIA as a company, you know, that's such a big part of what NVIDIA does.
There's this term that's sometimes used, general purpose technology.
What is that and why should that matter to government leaders or policymakers who are dealing with AI?
Yeah, it's general purpose technology is one of these terms that's very academic.
So it's probably not used that often, but it provides a really important perspective on why this is different.
Why are so many countries focused on this?
Why should we care? If you're looking for a very credible answer to that fundamental question of why in a complex world with many competing priorities, why should we prioritize AI over all these other technologies?
Here's your answer, right? is the most important technological force of our time.
I mean, that's full stop. I mean, that's an opinion perhaps, but what's not an opinion is that AI qualifies as a general purpose technology or GPT.
And you can look this up. It's on Wikipedia, like everything else.
And GPT is a very important designation because they do not come along that often.
This is a super rare thing, right? So electricity is a frequently cited example.
And And, and some, you know, it's often for AI isn't referred to as the sort of the new electricity.
That is not a bad metaphor. And here's why.
I mean, you know, electricity, the steam engine, information technology.
These are these share some common characteristics.
They may sound like, hey, well, that's all part of our lives today, not a big deal.
But actually, if you look at literally the history of mankind, it's important.
So their impact is pervasive. Every sector of the economy, our personal lives, our public sector, Everything is transformed.
Nothing is left unaffected. That's one of the important qualities here.
A really cool invention that transformed one or two industries doesn't cut it.
Another important quality is that GPTs improve over time.
So they become better, more efficient, more effective.
And I think it's safe to say the cost. and the safety and the mechanisms for delivering electricity have all improved greatly since the 1800s.
And Certainly, if you look at information technology more recently, as an example, you could see that trend as well.
And finally, GPTs are like a platform. They're not just impactful by themselves.
They spawn innovation. new products, new services, entire new industries are really born on the backs of GPTs.
You need to see that kind of impact. And when you add them all up, you have this list of innovations, right?
And to really grasp Noah, to really grasp this point, you have to go back, like way, way back.
There's only been 25 general purpose technologies since the dawn of time.
And I'm including things like the wheel, and the domestication of plants.
This is what it is. They're transformative, yeah.
You know, in a very defined way, AI is arguably the 26th general purpose technology. in 10,000 years.
So yeah, I'd say it's special and it probably deserves some policy attention.
Our guest today is Keith Stryer. Keith is NVIDIA's vice president of worldwide AI initiatives.
And we're talking about his work with countries around the world as they adopt and shape and reformulate their national AI strategies to really... prepare themselves or were already in it, but move forward in this world where AI is arguably the 26th general purpose technology we've ever seen.
Certainly, touching our lives in innumerable ways.
You know, Keith, it's one thing, and I know you well know this, but it's one thing to... create a strategy for something to write up a national policy document, even outlining an AI strategy.
You can go through the motions, you can get everybody aligned on their talking points and their slides, and you can publish it.
But where do you go from there? How do you go from strategy to action?
Is it as hard as I'm guessing it is? What's the process like?
Yeah, I mean, this is where the storyline gets more complicated.
First off... Many articles have portrayed this surge in activity around national AI plans as kind of a global arms race.
I don't agree with that. I don't think that's very productive, really.
I mean, no doubt the biggest are able to make the biggest investments in terms of dollars and they graduate the most data scientists write the biggest checks to the most number of AI startups and that kind of thing.
But this is about more than scale. You need big compute platforms.
But you also need people and talent and companies and policies to innovate on top of them.
You need the ecosystems and the programs that feed them.
You need that local communities and the networks of investors and students and and establish business leaders.
You have to create value through that connection.
You need the fiber in the ground. You need the data center above it.
You need the supercomputer, but you need the data to feed that machine.
So there isn't one component. And frankly, you might have the right mixture of those components, even in very small countries.
And that's why it's, It's important not to think of it as just a global arms race.
Let's look at Canada as an example. And they're a fraction of the size of the US, yet they lead the world. in one of the most important areas of AI, which is deep learning.
And this is the key discipline, the key area approach that really gave birth to the modern era of AI.
So many big tech firms, including Nvidia, have labs and outposts in Toronto or Montreal or Vancouver. to benefit from that rich talent pool up north and the infrastructure, in a sense, and that incredible community that has emerged up there.
Other countries like Singapore and Finland, and again, the UAE, these are countries without the conventional scale of large industrial countries, but nonetheless have emerged. as these lighthouse examples of how governments can access the broad benefits of AI and, frankly, make material contributions to the global economy through their example. government's getting involved in the right ways?
Or what is it about Canada, Singapore, UAE, Finland, these countries you mentioned that you know, maybe aren't the largest countries in the world by population, but are leading in these important areas of AI.
Do you have any insight into why them? Yeah, I think that leads to sort of a different question.
And so let me restate your point. Is there a common thread? that runs through countries that are clearly defining, gaining an edge.
So first I would say that there are an infinite number of use cases and applications, right?
There's so many subdomains and specialties.
I mean, AI is not a single technology. It's a constellation, right?
So when it comes to AI, there won't be one track, one trophy, you know, from research to applied fields to industry verticals, right? including regulation ethics.
I mean, there will be many winners and many opportunities for leadership.
But some countries, you know, Noah, some are stuck in sort of a committee form.
And so if you're looking for a pattern of success, you don't need a supercomputer to see it.
There's sort of the focus on policymaking and press releases, but lighter on action.
And the countries that want to see results need to step up on at least three fronts to gain an edge.
And this is just based on sort of having this backseat uh you know across the global economy to what's working right right so first first they need to shift their mindset and you said it even at the beginning uh at one point, right, kind of going from focusing on action.
They need to shift the mindset if they have it already from strategy to implementation.
They need to stop talking about AI and actually work on it.
A national AI strategy is never completed.
It's important to continuously revisit, challenge, update.
But citizens will see little impact without concrete action.
And that has to happen really on two levels, national agencies local agencies, including cities.
That's really the front lines. Now, there's three examples of action in that case.
One is sponsoring pilots. creating showcases within the public sector that inspire or lead the way for the private sector.
Another example of concrete action is really going out of your way to forge connections between industry, academia, and government.
And then also, frankly, governments need to use regulation to spur innovation. right?
Favorable tax policies and immigration policies that create the conditions for AI adoption.
And the immigration policies are important, particularly if you're talking about bringing in workers and students or professors from abroad.
So that's one area. The second area is scaling up investments in domestic AI infrastructure. those countries that have taken this concrete action to really understand that without world-class AI infrastructure, they can't be world-class in AI.
I mean, it's just that simple. And that same infrastructure supports digital sovereignty.
It drives economic growth. It drives job creation.
I mean, The same platform that serves physicists also serves entrepreneurs and small businesses.
You know, it's really about understanding that you need to make those investments.
And and that's we see that we see that again, countries large and small, the ones that have just moved aggressively towards that.
They just have an advantage. It's very hard to attract the world's top researchers.
It's very hard to keep entrepreneurs in your city. if they don't have the infrastructure they need to build out their vision.
But another key point is that AI infrastructure is not just about data centers.
And I think this is a technical nuance that gets lost in the conversation.
It's not just about big computers in big data centers.
That is a part of it, but that's not it.
AI happens everywhere. right, increasingly at the edge.
So you can have AI supercomputing in the form of a workstation that sits on the desk of a data scientist.
You can have it packaged up inside of the base tower of a 5G tower.
It can be at the corner of an intersection powering smart traffic management.
It could be in the back of a warehouse or at an airport gate.
So AI happens everywhere, including in the public cloud, with Amazon and Microsoft So you really have to think about supercomputing not as just big machines and big data centers but about this domestic capability that stretches across the public and private sector.
And from small machines to buildings, it's a very distributed sense.
And I don't think that was not the traditional way of thinking about it.
And there's one final area of action, which is the AI ecosystem.
And this is probably the most important.
I mean, if you have a great strategy and great infrastructure, but you don't layer on top of that the startups, the investors. all the other talent and programs that you need to create value from that.
It's hard to pull that together. And concrete policy action makes a difference.
I mean, you want to You want to attract and keep those professors and those students, right?
Who invent the AI architectures and algorithms.
You want to support the entrepreneurs that... that inspire young people to invent their own futures.
And you want to make sure you establish industries, the ones that provide a lot of the jobs today and the investors that support them.
You want to give them incentives to reinvest and opportunities to attract foreign investors and find new customers and all that stuff.
And maybe the fourth category right under that same bullet is just pioneering government leaders.
There are innovators everywhere, including in agencies and in ministries and in city authorities, pioneering government leaders that want to really show the way with improving public service and a better citizen experience.
And so the technology that helps the startup down the street is the same one that'll help an agency.
And I think it's important to see that connection.
So as we're recording this, it's the end of April 2020, and we're in the midst of the COVID-19 pandemic that's really gripped the in the entire globe in kind of an unprecedented way.
On the upside, there's been a lot of And I'm sure you're privy to more of this than most.
But even I've lifted my head out of the sand to look at the news a few times here and there and just seen- governments around the world and private sector companies stepping up and cooperating and working together to...
Yeah. Yeah. How has this changed the national AI policy landscape and investment priorities as well?
Yeah, it's a good question. I mean, I'm locked in my house with my three teenage children.
Exactly. And my four dogs. And it's, yeah, you can't deny that.
I have two kids and two goldfish and I'm struggling, but three teens and four dogs.
Yeah. We need a whole, we need a whole nother podcast.
So, yeah. So the world is battling COVID-19 on many fronts and certainly. there's a very strong intersection with this conversation.
Everyone's top priority, of course, is to flatten the curve and to keep citizens and healthcare workers safe.
And to stay grounded, and you just made this point, but it's worth restating.
I mean it's fair to say that as important as AI is, there's a bigger focus probably on finding solutions. masks and gloves and ventilators in the world.
And then so, you know, there's sort of that Maslow's hierarchy of needs.
Sure. But nevertheless, you Countries with more established domestic AI infrastructure, more mature AI ecosystems who already operate the most powerful AI supercomputers and have the best-in-class data governance and access to data.
These are the countries that are probably leading the world. in the development of treatments and vaccines.
And everyone will benefit, right? There's no question.
I think everything will be shared. But the leaders...
Frankly, I think there's a close relationship and correlation between having invested in that shift from just having an AI policy to building the programs, the infrastructure, the ecosystems, and then having the readiness and the preparedness today to deal with this.
In fact, Canada, as I mentioned, was the first country to do this, and they are a standout in the global AI map.
And as it turns out, this is anecdotal, but I think it's worth mentioning that the first detection of this outbreak, which really happened a few days before or a few weeks before it was really captured by the world, is widely credited to a small venture in Canada called Blue Dot.
Oh, wow. Okay. They were a startup that raised one of the earliest alarms about the risks posed by this virus that emerged in the Chinese city of Wuhan.
And of course, they're leveraging AI and data science to do that.
And so I just think that anecdotally, you can draw your own conclusions.
Yeah, well, I actually grew up about four hours or so south of Montreal in upstate New York, so I've always been a big fan of Canada.
Not surprised to hear that. But so all of that being said, you know, we usually like to end these conversations on a bit of a forward looking note.
So I'll just throw a big question at you then.
What's next? What's next for AI, national AI strategies and policies?
And just how does the world move forward with AI as part of the plan?
Well, you know, we all lived through the dot com boom, you know, 20 years ago and and, you know, in the rise of. these e-business groups and the e-everything.
Then eventually those capabilities didn't go away, they just blended in.
You didn't need an e-business team or an e-anything because everything became enabled by the internet eventually to some degree.
And I think there's a part of that's gonna happen here too.
I mean, national AI plans are important Most people don't understand what artificial intelligence is, its value, its limits, its risks, its potential across the public and private sector.
There's an important journey the countries need to go through and the leaders within those countries need to go through to really understand how best to position their country, to capitalize on the benefits, to to also manage the risks and all the above.
It plays an important role now. I wouldn't be surprised if 10 years from now, having a national AI strategy you know, wasn't really necessary for most economies because it had already happened.
And Basically, that strategy has been implemented and baked in, and it's now part of the new normal of how government is run.
So that's going to happen. But in terms of the most immediate impact, I'd say what's next is that the national AI frameworks that were developed, again, around sort of these generic areas.
I think some of the specific implementations we're seeing Obviously, there's a huge mobilization of resources around the global health crisis. and dealing with the specific healthcare issues and life science issues there in terms of developing treatments and vaccines and diagnostics.
And I think what's next is frankly leveraging AI and all those investments to really reboot economies.
And clearly that's the big task that we're gonna face in the next year or two.
So, and I think as countries slowly reactivate public life, trade and commerce, AI infrastructure will be a key enabler. of that.
And we're always starting to see that play out, frankly.
I mean, there are some great examples using computer vision to monitor social distancing and even whether people are wearing masks in public spaces, right? create sense of safety, using things like our virtual GPUs to improve the desktop experience, making it easier for people to work remotely and stay productive. and safe from home, modernizing the data centers, frankly, to operate more sustainably, to create efficiencies and hopefully pass on that cost to lower rent to tenants and boosting their ability to hire people back to work.
So I'd say net-net, the national AI plans will evolve.
And if done right and acted upon, there'll be much more than a policy document, which would be a shame, right?
The national AI plans really is a roadmap for the infrastructure, the new critical infrastructure that will propel countries forward. helping them rebuild their economies in a way that hopefully make them even more resilient to future shocks.
And, of course, NVIDIA and myself especially will be privileged to work side by side with many countries supporting them on this journey.
Well, it's quite a moment to be alive, to say the least.
And you're in the midst, in the thick of a lot of it.
So... We appreciate you taking the time to come on the podcast and certainly we covered a lot of ground, but for folks who want to know more about Your role, what NVIDIA is doing, the whole concept of national AI policy plans.
Is there somewhere online you could direct them to do a little more digging on their own?
Yeah, I mean, we have a great website. Both our NVIDIA blog, frankly, highlights many of our collaborations, like our recent...
AI Nations partnership with Italy, as well as our public sector section of the site.
And then frankly, you can just contact me directly and I'm happy to engage and share whatever people want to know.
Excellent. Well, Keith, thanks for taking time out from helping the world figure out its AI strategy and taking care of your family and your dogs.
Great conversation. And I'm sure it's one that folks will go back and re-listen to because there's a lot packed in there.
Stay well and best of luck to you on all the work you're doing.
Thanks, Noah. All the same to you. Bye.
