A primary goal of the National AI Initiative is for the United States to lead the world in the development and use of AI in the public and private sectors. Federal agencies are contributing significantly to the research, development, demonstration, and use of AI in a wide range of applications across society. These efforts are leading to breakthroughs in improved healthcare, safer and more efficient transportation, personalized education, significant scientific discoveries, improved manufacturing, increased agricultural crop yields, better weather forecasting, and much more. Today’s AI advancements are in large part due to decades of long-term Federal investments in fundamental and translational AI R&D, which have led to new theories and approaches for AI systems, as well as applied research that is allowing the translation of AI into practical applications.
Confronting one of the biggest challenges facing agriculture today—feeding an additional 2 billion people by 2050—the U.S. Department of Agriculture (USDA) is positioning farmers, scientists, educators, and the American public to benefit from artificial intelligence, providing economic opportunity through innovation, helping rural America to thrive, and promoting more efficient and profitable agricultural production.
In 2020, the USDA National Institute of Food and Agriculture (NIFA) launched two AI Research Institutes with the National Science Foundation, as well as a data science initiative, Data Science for Food and Agriculture Systems, to accelerate and expand on a diverse portfolio of AI-related programs that represent a multitude of uses in agricultural production, sensor development, bioinformatics, ecosystem management, rural community support, and workforce development through education and training at all levels. This work includes robotic solutions that utilize AI technologies to assist in pollination, weeding, pesticide applications, and fruit harvesting; AI algorithms that assist in identifying plant, animal and tree species that contribute to pest control and ecosystem management; and adaptive groundwater and watershed models to maintain resilience of agricultural systems. NIFA’s investments contribute to a wide breadth of AI-relevant research including big data, machine learning, autonomous systems, computer vision, and intelligent decision support systems, as well as the socioeconomic and workforce considerations that come with the rapidly increasing role of AI in U.S. agriculture.
USDA’s Agriculture Research Service (ARS) is collaborating with industry to advance the role of AI in monitoring livestock, using robots to sort harvests, analyzing water sustainability and pest management, and utilizing UAV technology to analyze crop health and efficiently apply pesticides. Their projects include crop sustainability, the use of automated calculations to analyze crop foliage composition and then guide the application of pesticides; and self-propelled apple sorting machines that use algorithms to quality sort the fruit. USDA ARS SCINet offers high performance computer clusters, cloud computing, and improved networking resources and training to researchers.
USDA’s Economic Research Service (ERS) is conducting research and development to use machine learning and AI to create better crop yield models based on weather data and analysis.
USDA’s work in AI and data science brings revolutionary technology to agriculture with the potential to transform our Nation’s ability to bring high-quality food to America’s dinner table.
The Department of the Treasury is pursuing policies that promote the adoption of innovative tools such as AI and machine learning to empower people to make more informed decisions about their short-term and long-term financial goals.
The U.S. Securities and Exchange Commission (SEC) is actively implementing machine learning algorithms to monitor and detect potential investment market misconduct. Additionally, the Consumer Financial Protection Bureau (CFPB) issued new policies that allow for an increased use of data and machine learning algorithms in financial products and services. These policies are helping unleash innovation in the financial sector, driving competition that lowers prices and provides consumers with more and better products and services. New approaches can also expand access to the benefits of the financial system for Americans of all backgrounds.
Financial institutions have been improving their ability to monitor transactions and conduct link analyses with new technologies that rely on artificial intelligence and machine learning. Being proactive on this front is more important than ever given the sophisticated approaches actors use to move money and goods. These efforts are using new technologies to identify and build out networks and help businesses and institutions make better financial decisions.
Artificial intelligence has immense potential to advance healthcare, accelerate medical research, and support the health and well-being of all Americans. This is especially true as AI technologies are leveraged as part of the global response to the coronavirus pandemic.
Beyond the pandemic, the Federal Government has taken important action to advance the use of AI innovation in healthcare. The U.S. Food and Drug Administration has a key role in this effort, starting with the first FDA approval of an AI-powered medical device in 2018. Since then, the FDA has permitted marketing of AI-based software that can help healthcare providers detect wrist fractures more quickly, and began developing an adaptive framework for smart software in medical devices, such as electrocardiogram (EKG) devices that estimate the probability of a heart attack. The FDA, along with the Centers for Disease Control and Prevention (CDC), partnered to advance research in machine learning and natural language processing by creating free tools to improve the collection of clinical data. In January 2021, FDA released the agency’s first action plan to advance oversight of AI and machine-learning-based medical software.
The Department of Health and Human Services (HHS) is promoting medical innovation through the Division of Research, Innovation, and Ventures (DRIVe) by offering partnerships with public and private institutions. For example, DRIVe seeks to provide innovative solutions to diseases such as sepsis by introducing machine learning algorithms to treat the disease. HHS also worked on a Health Tech Sprint to show how AI can be applied to Federal data to create products for healthcare applications.
The Centers for Medicare & Medicaid Services’ (CMS’) Center for Medicare and Medicaid Innovation (Innovation Center) launched the Artificial Intelligence (AI) Health Outcomes Challenge, in collaboration with non-profit and philanthropic organizations. The CMS AI Health Outcomes Challenge will distribute awards to encourage further progress in AI for health and health care and to accelerate development of real-world applications for this technology, such as predicting unplanned hospital and skilled nursing facility admissions and adverse events.
The National Institutes of Health is exploring many opportunities for AI to accelerate medical advances in biomedical research. The NIH Common Fund’s Bridge to Artificial Intelligence (Bridge2AI) program will propel biomedical research forward by setting the stage for widespread adoption of AI that tackles complex biomedical challenges beyond human intuition. NIH also has large data sets resulting from projects such as the NIH Human Microbiome Project and the All of Us Research Program. These data sets provide great opportunities for AI to foster discovery. In 2019, NIH reported that researchers were able to use AI to catch irregular heartbeats, improving the accuracy and efficiency of EKG readings. NIH also launched the Advisory Committee to the Director Working Group on AI to explore ways to harness the potential of AI to advance biomedical research and the practice of medicine. NIH is also working to reduce inequities in healthcare, increase diversity among AI researchers, and close gaps in health-related datasets. The NIH Artificial Intelligence/Machine Learning Consortium to Advance Health Equity and Researcher Diversity (AIM-AHEAD) program is building regional, multi-disciplinary partnerships to better engage underrepresented communities.
The Department of Veterans Affairs’ National AI Institute is working to build AI R&D capacity in order to improve the health and well-being of our Nation’s Veterans, including through pilots, processes, partnerships, and policies for AI. The VA also leads AI Tech Sprints that leverage a voluntary incentives framework to link the ecosystem of Federal, industry, academia, and nonprofit organizations around AI R&D for Veteran health and well-being. These efforts resulted in a number of AI-enabled tools, such as the Clinical Trial Selector that empowers 9+ million Veterans and 50+ million CMS beneficiaries to find clinical trials based on their own medical record.
The United States Agency for International Development (USAID) partnered with non-profit foundations to develop a report on AI in Global Health, which identifies opportunities for donors, governments, investors, the private sector, and other stakeholders to explore and accelerate the appropriate development and cost-effective use of AI at scale in global health.
COVID-19 Pandemic Response
credit: Jill Hemman/Oak Ridge National Laboratory, U.S. Department of Energy
Enlisting world-class science in the fight against COVID-19
Working alongside industry, academic, and nonprofit leaders, the Federal Government coordinated the release of the COVID-19 Open Research Dataset (CORD-19), the most comprehensive collection of machine-readable scholarly articles on coronavirus to date. The White House then issued a call to action to the AI community to develop publicly available AI tools and techniques that can help researchers summarize and analyze the dataset. The National Institutes of Health’s (NIH) National Library of Medicine adapted its standard procedure to ensure articles and accompanying data are available in machine-readable formats. The National Institute of Standards and Technology (NIST) and White House Office of Science Technology Policy launched a joint effort to develop search engines for research to fight COVID-19.
Capitalizing on America’s world-class AI and HPC capabilities, the White House led an effort with the Department of Energy, the National Science Foundation (NSF), and leading industry, academic, and international partners to establish the COVID-19 High Performance Computing Consortium. Through the consortium, Federal government, industry, and academic leaders volunteer free compute time and resources to COVID-19 researchers. Using consortium resources, academic researchers used machine learning and physics-based refinements to model high accuracy SARS-COV-2 membrane proteins, as well as to adapt a workflow designed for Ebola to design novel peptidic inhibitors of COVID-19 main protease.
NSF issued a Dear Colleague Letter on COVID-19 to solicit rapid proposals to explore how to model and understand the spread of COVID-19. NASA employed AI to predict the spread and impacts of COVID-19 and sponsored a hack-a-thon to develop COVID-19 solutions. Federal agencies, such as NSF and the Department of Energy’s (DOE) national labs, mobilized staff to use AI to accelerate, and screen drug discovery to fight COVID-19. The U.S. Department of Energy’s Oak Ridge National Laboratory used AI to accelerate COVID-19 treatment. Lawrence Berkeley National Lab used AI to estimate the seasonal cycle of COVID-19, map its proteins, and used natural language processing to draw previously undiscovered insights and connections. The NIH harnessed AI’s image recognition abilities to diagnose, treat and monitor COVID-19 from medical imagery.
The Department of Defense’s Joint AI Center’s Project Salus used AI to integrate supply chain data to improve COVID-19 crisis response. The Department of Justice in cooperation with other agencies and private-sector companies used cyber enforcement to disrupt hundreds of on-line COVID-19 related scams.
In these and many other ways, the Federal agencies are leveraging the power of AI to combat the COVID-19 pandemic.
National Security and Defense
The DoD AI Strategy focuses efforts on harnessing AI to advance our Nation’s security and prosperity. The DoD AI Strategy defines the Joint Artificial Intelligence Center (JAIC) as the focal point of DoD’s AI efforts, and outlines the following key strategic aims: delivering AI-enabled capabilities for key missions; partnering with leading private sector technology companies, academia, and global allies; cultivating a leading AI workforce; and leading in military ethics and AI safety. The JAIC aims to spur momentum in the use of AI for DoD by focusing on a set of challenging use cases that can benefit from AI, including perception, predictive maintenance, humanitarian assistance and disaster relief, and cyber sensemaking. The JAIC’s mission is to both deliver new AI-enabled capabilities to DoD end users, as well as to incrementally develop a common foundation of shared data, reusable tools, frameworks, libraries, and standards that are essential for scaling the impact of AI across DoD.
The work of the JAIC is also benefitting other applications beyond defense, such as natural disaster response. As a spin-off of JAIC efforts in humanitarian assistance and disaster relief, the First Five Consortium was formed as a collaboration across government, industry, and academia to mitigate the impacts of natural disaster through the application of technology. The Consortium aims to put technology into the hands of first responders to enable them to more effectively respond to disasters.
The Defense Advanced Research Projects Agency’s AI Next Campaign is a multi-year investment of more than $2 billion in new and existing programs. Key areas of the campaign include automating critical DoD business processes; improving the robustness and reliability of AI systems; enhancing the security and resiliency of machine learning and AI technologies; reducing power, data, and performance inefficiencies; and pioneering the next generation of AI algorithms and applications, such as explainability and common sense reasoning.
In the intelligence community (IC), the Augmenting Intelligence using Machines (AIM) Initiative outlines how the IC is incorporating AI, process automation, and IC officer augmentation to improve mission success and efficient. With scale in mind, the IC is enhancing its ability to provide much-needed data interpretation to decision makers across government. The Intelligence Advanced Research Projects Activity (IARPA) sponsors a number of programs that leverage or improve AI and machine learning for applications such as video search, enhanced text retrieval, detection of cyber-attacks, event forecasting, and related topics.
The National Security Commission on AI released its final report in March 2021, concluding that significant action is needed “to accelerate AI innovation to benefit the United States and to defend against the malign uses of AI”. The Commission identified four key pillars for immediate action: leadership, talent, hardware, and innovation investment. The report provides recommendations for responsibly employing AI for national security, and for actions the U.S. government should take to promote AI innovation to improve national competitiveness and protect critical U.S. advantages. The report emphasizes the importance of working in partnership with industry, academia, civil society, and democratic allies in order to compete and win the values competition inherent in the use of AI.
The Department of Homeland Security’s (DHS) AI Strategy aims to enhance the capability of DHS to safeguard the American people, the homeland, and national values through the responsible integration of AI into the Department’s activities and by mitigating risks posed by AI. Through the appropriate use of AI, DHS will take advantage of new opportunities to secure the Nation, identify and interdict criminal actors, and secure cyberspace.
In the last two decades, AI has helped accelerate scientific discoveries in numerous domains. The Department of Energy National Laboratories, including Argonne National Laboratory (ANL) and Oak Ridge National Laboratory (ORNL), are leading numerous efforts in the use of AI to advance science. The ability of AI to process big data and automate repetitive tasks has sped its adoption in many scientific fields. For example, AI is advancing materials science and drug discovery. AI is being used to match cancer patients with clinical trials. Medical diagnostics, decision making, and bio-engineering designs have benefited from AI algorithms. AI is being used to enhance human decision making in complex systems, and in astronomy to find and detect new phenomena. An extensive report on AI for Science produced by ANL, ORNL, and Lawrence Berkeley National Laboratory outlines many other ways AI is advancing science. Additional R&D into how AI can be employed in scientific endeavors will further increase the pace and quality of transformative discoveries.
Autonomous systems, such as unmanned aircraft systems (drones), Urban Air Mobility (UAM), Advanced Air Mobility (AAM), and self-driving vehicles, offer tremendous benefits to our economy and society. They promise to transform the delivery of household goods, provide mobility options for senior citizens and Americans with disabilities, improve the safety of dangerous occupations, and expand access to life-saving medical supplies. In partnership with State and local governments, the Federal government, through the Department of Transportation (DOT) and NASA, is working to enable the safe operation of these systems in our airspace and on our roadways.
In 2017, the Federal Aviation Administration (FAA) began the Unmanned Aircraft System (UAS) integration Pilot Program (IPP), which tested and evaluated the integration of civil and public drone operations into national airspace in collaboration with State, local, and tribal governments. Participants in this program explored concepts such as night operations, flights over people, beyond line of sight, package delivery, and other operations. The FAA concluded the IPP in October 2020 as mandated by statute. This program helped DOT and FAA define new rules that support more complex low-altitude drone operations.
To further the integration of UAS into our airspace, the FAA has defined new rules for remote identification of unmanned aircraft and for the operation of drones over people. These new rules are the key to unlocking beyond visual line of sight (BVLOS) operations. To that end, the FAA created the BEYOND program to address the remaining challenges of UAS integration. BEYOND focuses on expanding BVLOS operations, collaborating with industry, and community engagement to permit flight operations under established rules rather than waivers. Further, the FAA has established the UAS Data Exchange, a collaborative approach to facilitate the sharing of airspace data between government and industry, which has already led to shortened processing times for airspace authorizations for UAS operators. In coordination with the FAA, NASA continues development of a UAS Traffic Management (UTM) system to enable safe, efficient, and equitable small UAS operations at scale. NASA’s Aeronautics Research Mission Directorate and Agility Prime, a United States Air Force program, are working to safely develop AAM vehicles (i.e., “flying cars”). These aircraft will incorporate non-traditional electric or hybrid electric propulsion.
The development and deployment of automated vehicles (AV) and automated driving systems (ADS) have the potential to reduce the number and severity of serious automobile crashes. A high percentage of these crashes are due to human error, according to the National Highway Traffic Safety Administration (NHTSA), and automation can help prevent injuries and save lives. The market for automated vehicles is anticipated to continue to grow. In addition, fully automated vehicles have the potential to present new transportation options for older Americans and those with disabilities, increasing their connectivity and independence.
DOT is taking active steps to develop guidance for how best to integrate automated vehicles and driving systems into our transportation system. In January 2020, DOT released “Ensuring American Leadership in Automated Vehicle Technologies: Automated Vehicles 4.0,” which details ten U.S. Government principles to protect users and communities, to promote efficient markets, and to facilitate coordinated efforts to ensure a standardized Federal approach to American leadership. This new document builds off of the Department’s 2018 AV 3.0 guidance and 2017 ADS 2.0 guidance, which focused on innovation for surface transportation modes, and safe testing and integration of Automated Driving Systems, respectively.
DOT’s Intelligent Transportation Systems Joint Program Office has also identified practical real-world scenarios where AI offers the potential to address specific transportation needs. They have also identified broad categories of AI-enabled applications that can be applied to address specific transportation problems and needs and summarized existing and potential transportation applications enabled by AI under each category.
credit: Integrated Remote and In-Situ Sensing (IRISS), CU Boulder
Drone approaches supercell thunderstorm, helping scientists better understand and forecast tornadoes
The National Oceanic and Atmospheric Administration (NOAA) is using AI in every mission area to better understand and predict the dynamic environment in which we live. NOAA’s AI Strategy is advancing AI research and innovation in NOAA missions and accelerating the transition of AI research to applications. NOAA is prioritizing R&D investments that will accelerate the ability of researchers to create AI that can assimilate the massive amounts of big data from environmental satellites into weather models to improve predictions of hurricanes and severe storms. AI capabilities, including sensors on aerial and underwater drones, are helping NOAA to improve weather forecasts and understand derechos and flooding.
NOAA’s additional applications of AI include improving nautical charts to ensure safe and efficient maritime commerce, surveying fish stocks to effectively manage our nation’s $208B/year recreational and commercial fishing industries, monitoring and conserving endangered species, and exploring, mapping, and monitoring the world’s ocean conditions and coasts for critical national security and economic applications. NOAA also uses onboard AI and machine learning techniques in NOAA satellites help protect the environmental spacecraft during radiation events that could corrupt satellite computers. Automated software checks built into the onboard system can detect charged particles passing through onboard electronics, and then initiate contingency procedures to clear error conditions and help return satellites to full functionality.