Among the many departments and agencies within the United States federal government, the US Department of Energy (DOE) stands out as one of the most science, technology, and innovation-focused. This should come as little surprise to those who know the DOE’s storied history with its breakthrough labs, world-leading research institutions, and highly educated staff. Since World War II, the DOE has been at the forefront of most of the groundbreaking and world-changing revolutions in science and technology including the development and harnessing of nuclear energy, innovations in genomics including the DOE initiative Human Genome Project, work in high-performance computing, and many other research-oriented efforts.
In fact, the DOE supports more research in the physical sciences than any other US federal agency, providing more than 40% of US funding in computing, physics, chemistry, materials science, and other area through a system of national laboratories including Lawrence Berkeley National Laboratory, Oak Ridge National Laboratory, Argonne National Laboratory, Ames Laboratory, Brookhaven National Laboratory, Los Alamos National Laboratory, Sandia National Labs, Lawrence Livermore National Laboratory, the SLAC National Accelerator Laboratory, and dozens more institutions. Until very recently, the DOE also ran the world’s top two fastest supercomputers: Summit and Sierra.
With all this research, computing, and funding strength, it should come as little surprise that the DOE is spearheading the charge to advance research into AI and its applications across a wide range of industries and uses. Earlier this year, Cheryl Ingstad was named the Director of the AI and Technology Office at U.S. Department of Energy (DOE). She will be speaking at an upcoming AI in Government event in August 2020, and in this article, she shares some of the insights she’ll be sharing with attendees.
People might not know that the DOE has a long history with AI. What is the history of AI at DOE?
Cheryl: With its near 80-year legacy of innovation, unmatched R&D infrastructure, and unquenchable thirst for accelerating scientific discovery, I would argue that DOE and our 17 national labs, are the single biggest incubator of innovation that our country – and the world – has ever known.
Our history in AI goes back to the 70’s and 80’s with rule-based and early machine learning approaches for operating large experimental facilities, like particle accelerators, and analyzing large data sets.
In the 1990’s and 2000’s there was extensive work at the labs in applying emerging machine learning methods to high-consequence national security challenges, like cybersecurity and nuclear nonproliferation, that required well-understood performance and confidence evaluations. Work on scaling up learning systems on DOE supercomputers also started in this time frame.
Suffice it to say, our enterprise has been hard at work – often in partnership with the private sector and academia – in machine learning and artificial intelligence for decades. Because of that record, and its unrivaled computing capabilities, DOE is uniquely positioned to lead the Federal Government’s efforts to develop, advance and deliver on the promise of artificial intelligence.
The DOE has been using AI and ML in some very advanced ways. Can you share some examples of how you are applying AI?
Cheryl: Today, DOE and its national laboratories are leading more than 600 different AI projects that are designed to strengthen our core missions of energy, cyber, and national security, and to accelerate scientific discovery.
We’re using AI to increase precision subsurface energy exploration, which not only increases production, but benefits the environment. AI is helping us boost the reliability and resiliency of our electrical grid, which provides the foundation for our economic and national security. We’re also using it to enhance our protections against cyberattacks on our critical infrastructure. These are all areas where you might expect DOE to be using this breakthrough technology.
But AI’s impact will extend well beyond DOE’s traditionally understood core missions, so our labs are also using it to boost crop yields, help predict seismic activity like earthquakes, accelerate the development of new medicines, and to improve treatments and cures for our most deadly diseases. Most recently, DOE has brought the power of Summit, [ed: until recently] the world’s fastest supercomputer, and our scientific and AI expertise to the fight against COVID-19.
As the first Director of the Artificial Intelligence & Technology Office (AITO) within the Department of Energy (DOE) how do you plan on developing, delivering and managing AI technologies in support of DOE?
Cheryl: First of all, let me say that it is an honor to have been appointed as AITO’s inaugural Director at a time of monumental importance for AI in a Department widely considered as the leader of civilian AI in the US Government. There is a period of great transformative change ahead of us and AI will be critical to our success.
As I mentioned, there are literally hundreds of promising projects touching nearly every corner of our vast enterprise, but until now, those efforts have been siloed, lacked coordination and strategic direction.
That is why this past September, former Secretary Perry established AITO to serve as the enterprise’s nerve center for AI work, to help assess, coordinate, drive DOE’s unmatched progress in this critical area.
The goal of AITO is simple: to organize DOE’s varied AI activities, whether they be R&D or applications or policy or infrastructure efforts, identify resources to accelerate their success, and most importantly, align them and focus them like a laser on ensuring AI is used as a force for good.
As Director, my priority will be to establish the AITO as the driver for AI transformation in DOE. This transformation will not be easy. There will be setbacks, roadblocks and challenges to overcome, but I am confident that working together, we can use AI to enhance and have an enduring positive impact on America’s energy, economic and national security.
What do you see as some of the unique opportunities the public sector has around AI?
Cheryl: What is so exciting about this technology is that the applications, potential optimizations, and benefits of AI are as broad as the imagination. Internally, at the direction of Secretary Brouillette, DOE program offices and labs across the spectrum are working in partnership with AITO to identify processes, from operations to procurement, to that are ripe for short-term progress. As the Secretary has said, it is essential that our enterprise leads by example in the demonstration of the benefits of AI.
For example, many people don’t know that DOE owns and operates a portion of the country’s electrical grid that roughly ten percent of the American population depends upon for reliable and affordable power. So, one of the tangible opportunities we have is to develop AI solutions and apply them to improve services and demonstrate a cost savings to our customers.
On the interagency front, DOE is working in partnership with several agencies on ways that AI can be applied to their missions with the near-term goal of being better stewards of taxpayer dollars and enhancing the services the federal government provides to all Americans.
But more specifically, another unique opportunity we have is in the area of secure storage, curation and analysis of critical government datasets. Every agency in the federal government generates massive amounts of data that are currently being underutilized. DOE has the secure enclave infrastructure, raw computing power and vast scientific expertise to serve as a clearinghouse and curator of those datasets to ensure the government is applying AI to the Nation’s toughest challenges.
What are some of the unique challenges the public sector has around AI?
Cheryl: As with any disruptive technological breakthrough, there are significant challenges to the status quo and legitimate concerns that need to be addressed.
They involve questions over whether AI-supported decisions can be verified, trusted and free of human biases. They involve legitimate concerns over data security, ethics and privacy.
There are real workforce challenges, not only about the eventual automation of functions traditionally performed by human beings, but how we help prepare people for the jobs that AI’s adoption will create. And there’s one more issue that is increasingly on people’s minds as nation states and non-state actors advance in their AI capabilities, and that is Adversarial AI.
These concerns are legitimate and the ramifications of these breakthrough technologies are real. The answers to these challenges will require as much innovation and imagination as the technologies themselves we are working so earnestly to develop.
How is the DOE working towards getting their data in a usable state for AI/ML?
Cheryl: That’s a great question. As you know, data is vital to the success of AI. As amazing as this technology is, it is only as reliable as the data you feed it. Recognizing this, DOE’s senior leadership has been working internally through its Data Governance Board, of which I am a member, to make best use of the vast amounts of legacy data we already have, and to ensure that data we gather going forward, is curated, labeled and secured in a way that enables timely and verifiable AI solutions.
One of the goals we have is to make critical data sets more broadly and easily available to our researchers in the labs. We want to reduce the amount of time our subject matter experts spend on data acquisition and data management so that we can accelerate the successful application of AI.
The DOE has the world’s fastest supercomputers. How are they being applied to AI?
Cheryl: Today, DOE’s computing capabilities, and our thirst for discovery, are second to none.
We house two of the world’s three fastest and most powerful supercomputers, and we will soon deploy three next-generation exascale machines, that will help us process astonishing amounts of data with extraordinary precision at lightning speed.
When AI insights are needed most urgently, as has become evident in the federal government’s response to COVID-19, America, and indeed the world, look to DOE’s supercomputers to accomplish in minutes and hours what would take other systems weeks or months.
In fact, DOE is one of the key partners supporting the COVID-19 High Performance Computing Consortium with our compute capabilities. This partnership is already utilizing AI to analyze massive data sets to discover novel insights regarding the coronavirus.
How important is AI to the DOE’s vision of the future?
Cheryl: I am convinced that AI will prove to be every bit as transformative to our daily lives as the onset of electricity was more than a century ago. Secretary Brouillette recognizes this as well. His vision is to transform the Department of Energy into the United States Government’s lead agency in the civilian use of AI by accelerating the research, development, delivery, and application of the technology.
AI is already supporting many of DOE’s key mission areas, including America’s energy reliability, environmental stewardship, nuclear security, national intelligence, and scientific discovery. It is also changing how we respond to natural disasters, safeguard our critical infrastructure, inform the missions of the intelligence community, and even how we detect and combat malicious cyber threats.
DOE has responsibility for high-consequence national security decisions that have little room for failure. Implementing safe and trustworthy AI solutions into many of these critical areas will require DOE to push the boundaries of currently understood AI frontiers.
What is the DOE’s perspective on ethics and responsible use of AI?
Cheryl: As the President’s Executive Order on Artificial Intelligence makes clear, it is absolutely essential that America leads the way in the development and application of AI because we must ensure that, as it becomes more prevalent in our daily lives, AI reflects our democratic values and respect for human rights and dignity.
Make no mistake, America is engaged in a new “space race” in AI and other industries of the future against countries that don’t share those values and have no interest in using AI as a tool to improve daily lives, but rather, as a weapon to control them.
America has the power and the obligation to lead in technological development, and to partner with the private sector, academia and our like-minded allies to shape the future of AI as a force for good.
What is the DOE doing to develop an AI ready workforce?
Cheryl: AI is creating entirely new career paths, while transforming others. Every sector of our economy is wrestling with this challenge, albeit from different perspectives. At DOE, enhancing the AI capabilities of today’s workforce, while recruiting for tomorrow, will be crucial to our ability to make the most of this technology to fulfill and optimize our missions.
To that end, we have begun the process of developing online training curricula and educational seminars designed to upskill nearly every level of our existing federal workforce. These modules, which we are developing with input from our labs, and our partners in the private sector and higher education, will demystify AI to the layperson, encourage employees to think creatively about how AI solutions can help them do their jobs, and empower them to proactively address the ethical, legal, and societal issues posed by the application of AI technologies.
An AI-ready DOE workforce must include AI technical experts to design, develop and benchmark AI technologies. So, we have a parallel strategy to develop, recruit, and retain a broad range of AI talent and expertise through internships, scholarships, fellowships, exchange opportunities, and visiting professor programs within our national laboratories.
But beyond DOE, I would argue that a comprehensive education initiative equal to that of STEM needs to be led at the national level to drive interest in AI-related courses and properly phased curricula must be developed to ensure the workforce of the future is educated and trained for careers in this groundbreaking field.
How is the DOE engaging industry and private sector in your AI efforts?
Cheryl: Our office has recently issued its first Request for Information (RFI) that asks the AI Community to contribute their ideas on what are the most consequential AI technologies and applications that will ensure American AI leadership in the next decade. I am eager to read the responses and expect that their insights and expertise will contribute to the direction of AI Grand Challenges in the near future.
What AI technologies are you most looking forward to in the coming years?
Cheryl: I am most excited to see how we develop and weave together multiple AI technologies such as neural networks, natural language processing, and making sense of sparse data, among others, to create solutions that transform our national security, businesses, industries, and health care. And personally, regarding health care, I view AI’s potential to save lives and treat our deadliest diseases as nearly limitless.