Skip to main content
Email Print Share

Remarks

Photo of Dr. France A. Córdova

Photo by NSF/
Stephen Voss

Dr. France A. Córdova
Director
National Science Foundation

Biography

At the
NVIDIA GPU Conference
Washington, DC

October 26, 2016

If you're interested in reproducing any of the slides, please contact the Office of Legislative and Public Affairs: (703) 292-8070.

[Slide #1: Advancing the Frontiers of Science]

Good morning, welcome to Washington, DC. It's a pleasure to be with you.

Advances in information technology over the last half-century have transformed our lives in ways too numerous to recount. And yet the IT changes we see unfolding today will have an even greater impact in the future.

One way that is highly anticipated is through Artificial Intelligence (AI).

But before I go into depth about AI, let me tell you a bit about the National Science Foundation.

NSF was created in 1950 and is the only U.S. government agency dedicated to supporting basic research and education in all fields of science and engineering.

NSF works every day to support basic -- or fundamental -- research that engages the scientific curiosity of hundreds of thousands of scientists, engineers, educators and students across the country.

Let's go to the video to hear more.

[Slide #2: Image unavailable]

As you just saw, NSF-supported research has had tremendous real-world impact. We enrich the lives of many millions of people around the world - including each of your lives!

In fact, NSF-funded research created the foundation of knowledge for many of today's innovations in GPUs and in AI.

[Slide #3: NSF by the Numbers]

NSF operates with an annual budget that is currently about $7.5 billion, and the vast majority of that - 93 percent - goes to support research, STEM education, and development of the STEM workforce.

The Foundation's annual budget represents just four percent of the total federal budget for research and development, but accounts for one quarter of the total federal support for basic research conducted at U.S. colleges and universities.

In many fields, NSF is the primary source of federal academic support. For example, in computing, NSF accounts for 82 percent of such investment.

We fund curiosity-driven ideas that push the frontiers of knowledge and discovery. These gains in knowledge have led to innovations with tremendous impact - for example, solar panels, 3-D printing, the Internet, lifesaving technologies and therapies, and much more.

[Slide #4: NSF is where all interesting research gets started]

As Eric Schmidt, former CEO of Google and current executive chairman of Alphabet, has said, NSF is "where all interesting research gets started."

This has been largely true for the field of AI. Consider self-driving cars - technology that's becoming more ubiquitous today and made possible with years of NSF funding. Advances in areas such as precision sensors; computer vision, planning and reasoning; real-time data analytics and predictive modeling; and control and automation have been integral in enabling the rise of autonomous vehicle technology. NSF has funded advances in all of these areas.

[Slide #5: AI at your fingertips]

You have only to address your smart phone to get a sense of where AI is and where it is headed. And, yes, NSF-funded research that has helped to enable that too! Advances in natural language processing and speech recognition technology have helped to put AI at your fingertips.

You can verbally ask for directions, and your phone can respond with the location of the nearest gas station or restaurant.

So, I asked Siri on my iPhone if she is intelligent.

She responded in an even voice, "You will have to judge for yourself."

Can you compose a song? She shuffled a play list.

Siri, can you dance? "Hmm. I'd rather sit this one out."

Just give her time...

[Slide #6: Enabling personalization]

AI is increasingly more common all around us and is evident in today's trend towards personalized everything.

There are personal tutors to help you with math or learning a new language; personalized, in-home robots to help as care givers and even to clean floors, and personalized medicine tailoring a treatment plan just for you.

To ensure that we reap the benefits of personalized everything - and of today's AI revolution more generally - we need to be sure that we consider the social and policy implications.

AI is a frontier where both openness and security are required but where the innovation that is needed must be done in concert with changes in policy and law.

If ever there were a definition of a compelling, yet challenging, "frontier," AI is it.

What excites me as Director of NSF is that the discovery space for AI is immense -- it has the potential to revolutionize everything we do.

[Slide #7: Look, Mom, I'm on the internet!]

I am reminded of the late 1980s and early 1990s, when I was a professor at Penn State.

Here I am!

Email was mostly used only among scientists, and there were no websites as we know them today. At the time, we couldn't have anticipated that the digital realm would explode and that our world would never be the same. AI has the same explosive potential.

It has the potential to enrich our lives. It can help us enhance our performance and augment our human capabilities.

But to get there, we need more AI research. By expert admission, we - government, academia, and industry - are "under-investing" in AI.

It's a point often made by Jason Furman, Chairman of the Council of Economic advisers - from whom you'll hear after me. He's observed, "The biggest worry I have is that we do not have enough of AI."

NSF agrees.

[Slide #8: Three AI reports]

And, this is echoed in 3 recent reports.

"AI-100" is the first report of the "One-Hundred-Year Study on AI" and its influences. This report, published in September by Stanford University, makes the point that "Throughout history, humans have both shaped and adapted to new technologies."

Indeed, computing technologies and human societies are co-evolving, transforming each other in the process.

The other two reports were released just two weeks ago by the White House, in conjunction with the Frontiers Conference hosted by President Obama.

One is a strategic plan and the other a vision document. These reports address the question, "How can America best prepare for an AI future?" They discuss the role of government and lay out a number of strategies and recommendations.

Together, these reports cover the landscape of AI from application sectors to potential impacts - positive and negative.

The White House reports recognize that Federal investments are imperative to ensuring that AI is brought to bear on challenges facing society today - to help improve the lives of people across the Nation and the world.

[Slide #9: NSF's unique role in supporting AI R&D]

NSF and fellow science agencies like NIH, NIST, DOE and DoD have a special role in supporting AI research. As you can see on this figure, our investments in AI generally support long-term, high-risk, fundamental research.

Industry often funds research over shorter time horizons that is lower risk and more "applied" with high immediate market-relevance.

NSF will fund long-term investments in AI research; enable effective methods for human-AI collaboration; and develop talent for the AI R&D workforce.

We will support researchers, developers, social scientists, and policy makers to seize the opportunities afforded by AI and to create the knowledge base to help overcome its associated challenges.

And, together, the efforts of government, academia, and industry will help the U.S. to maintain world leadership in AI.

[Slide #10: Hot areas of research]

There are a number of "hot" areas, ripe for increased research in AI.

For instance:

  • Large-scale machine learning - we've already seen computers reroute traffic because of busy intersections, and diagnose illnesses of patients. Can a computer address some of our questions about the universe, like the nature of 95% of its mass-energy content? As an astrophysicist, I'd like to know!
  • Deep learning - can a computer interpret a scene enough to recognize when a crime is being committed?
  • Apprenticeship learning - can AI teach robots to perform physical tasks?
  • Natural language processing - well, I'm still struggling with my virtual airline agent when I call for reservations - improved speech recognition could help here.
  • And of course, interdisciplinary studies of the societal impacts of AI, such as safety, privacy, and fairness, are important.

The research landscape of AI is broad, and NSF is invested in exploring it.

Not only does NSF support research that advances nearly all areas of artificial intelligence, we are also investing in the foundational areas that are enabling AI technologies.

We support basic research that helps to advance hardware and software.

Just think about the ways in which advances in GPUs alone have helped to democratize computing and, in the process, open the door to a new AI revolution.

And we are also investing in the research areas that are driving the AI revolution: from big data to more powerful systems capable of performing data- and computationally-intense calculations.

[Slide #11: Looking Ahead: Ten Big Ideas]

Building on these trends and continuing to expand the frontiers of science and engineering, NSF recently identified "Ten Big Ideas." These are bold ideas for the future -- areas that NSF is uniquely suited to address.

I'll highlight just a few of those where AI closely aligns.

[Slide #12: Shaping the Human-Technology Frontier]

The first I want to mention is Shaping the Human-Technology Frontier. Our current focus in this Big Idea is on the workplace.

Society today is on the cusp of a major transformation in work that is being driven by combinations of machine learning, artificial intelligence, the internet-of-things, robotics, and more.

AI is gradually shifting towards intelligent systems that work WITH people, not in place of them.

Indeed, for the last five years, NSF has led the National Robotics Initiative, investing in collaborative robots - what we call "co-robots" - that work WITH people on a variety of tasks and in a variety of settings.

Think about your own businesses: can you imagine how AI could just about 'do it all'?

Some jobs may be eliminated, but several studies suggest that many new ones will also be created. And without question, many jobs will require a tight interaction between people and computation, broadly construed.

NSF aims to catalyze interdisciplinary research as well as the education necessary to understand the benefits and risks of new technologies as we prepare for the future of work and productivity.

[Slide #13: Growing convergence research at NSF]

A second Big Idea is a new emphasis on Growing Convergent Research.

We recognize that it takes all kinds of scientists to address some of today's grand challenges.

We need physicists, computational biologists, software engineers, programmers, statisticians, linguists, and cognitive scientists, among others.

AI is inherently interdisciplinary. I know that many of you in this room are already using convergent approaches to develop exciting, new technologies.

Let me give you one example: Recently, a convergence team of doctors, theoretical physicists, and computational biologists - supported by NSF -- worked together to make a potentially ground-breaking discovery for pancreatic cancer. They found that mice with this cancer, when treated with a combination of chemotherapy and a synthetic form of vitamin D, lived 50% longer than those given chemo alone.

These results supported the theoretician's conjectures that synthetic vitamin D could help deactivate a protective barrier around the cancer cells and enable more chemo to reach the cells.

Obviously, that's an oversimplification of highly technical research, but it illustrates what can be done through convergent research.

[Slide #14: The Quantum Leap]

A third Big Idea is Leading the Next Quantum Revolution.

Building on decades of NSF-supported research, we are uniquely situated to address fundamental questions about quantum behavior and manipulation of quantum systems.

This effort comes at a time when we're approaching the limits of Moore's Law and in turn seeing new paradigms of computing emerge.

Exploiting inherently quantum phenomena - such as superposition, entanglement and squeezing - can help build more powerful computers, create exquisitely sensitive detectors, simulate physical systems, and enable new, secure ways of communicating.

Quantum computing, for example, will require continued research on algorithms, programming languages, and compilers. These approaches may help deliver new technologies for science, commerce, and defense.

[Slide #15: Harnessing Data ]

The next Big Idea identified by NSF that I want to mention is Harnessing the Data Revolution.

AI feeds on data - think about Watson, which outcompeted its human opponents in the Jeopardy! game show by quickly sifting through copious amounts of data.

Could Watson help us figure out intractable problems, such as the makeup of dark matter and dark energy, by quickly processing and synthesizing the large amount of information and simulations already available?

Is the answer already lurking in the vast trove of data, in publications of various languages?

[Slide #16: NSF-INCLUDES]

The last Big Idea I want to leave you with is NSF-INCLUDES. This idea aims to transform education and career pathways to help broaden participation in science and engineering.

NSF has long supported efforts to build a more diverse, inclusive workforce for all areas of science and engineering. In fact, we built the knowledge base and are one of the lead federal agencies for the White House "Computer Science for All" initiative.

CS for All is designed to bring computer science education to all K-12 students across the nation, with access and equity at the core. NSF has been a leader in encouraging rigorous and engaging computer science education in our nation's schools.

Bringing computer science to all students at an early age will help to excite and empower them to pursue the field, and AI, in their careers.

And we need that!

The White House Strategic Plan on AI R&D notes that, "62% of companies will use AI technologies by 2018, up from 30% today. [It goes on to say that] AI experts are already in short supply, with demand expected to continue to escalate... universities and industries are in a battle to recruit and retain AI talent."

[Slide #17: Looking to the future]

To summarize: We are well poised to embrace today's AI revolution. We need continued basic research to pursue the rich opportunities, and challenges, presented by AI.

Doing so will help to ensure that the U.S. remains a global leader; it will benefit the National economy, health, and defense; and it will enhance our lives -augmenting our ability to learn, to be more productive, and address society's largest challenges.

And, by working together - across academia, industry and government - we can ensure that everyone can enjoy the benefits of AI.

That's a Very Big Idea!