National Security Commission on Artificial Intelligence. Final Report

CHAPTER 15

The United States cannot compete with and counter the global

technology ambitions of authoritarian rivals if it acts alone. Like-

minded countries must work together to advance an international

rules-based order, protect free and open societies, and unleash

economic innovation. The authoritarian challenge to the global

technology order encompasses five distinct but related elements:

• A rising challenge to U.S. and Western technology firms for global market share,

impacting the prosperity and global economic position of the United States and its allies

and partners;

• China’s increasing influence and strategic leverage over countries that utilize

technologies and infrastructure built and developed in China;

• The prospect of authoritarian consolidation in states that gain easy access to digital

tools that can strengthen repressive rule;

• The prospect of democratic backsliding in states with governments that may be tempted

to utilize digital tools in ways that undermine liberal values; and

• A threat to the cohesion of democratic allies as an influential bloc of states with the

capacity to shape global technology norms and standards.¹

“The United States must

pursue a comprehensive

strategy in close coordination

with our allies and partners for

AI innovation and adoption that

promotes values critical to free

and open societies.”

243

A FAVORABLE INTERNATIONAL TECHNOLOGY ORDER

The United States must pursue a comprehensive strategy in close coordination with our

allies and partners for artificial intelligence (AI) innovation and adoption that promotes

values critical to free and open societies. Furthermore, the United States must collaborate

with its closest allies and partners to develop principles for employing AI tools ethically

and responsibly, defend the integrity of international technical standards, promote digital

markets, leverage comparative expertise to develop privacy-preserving technologies,

and share practices and resources to defend against authoritarian attacks on digital

infrastructure and democratic values.

To achieve these goals, the Commission proposes that the White House request the

Department of State to lead an effort with and other key agencies to:

Develop and implement an International Science and Technology Strategy (ISTS) to help

Recommendation

coordinate AI and emerging-technology policies government-wide and with our closest

allies and partners; apply the tools of foreign assistance, technical expertise and guidance,

and development finance; and foster collaborative R&D. The ISTS should serve as the

international component of the National Technology Strategy (see Chapter 9 of this report).

The ISTS should be centered around four big initiatives:

• Build an Emerging Technology Coalition of allies and partners to promote the design,

development, and use of emerging technologies according to democratic norms

and values; coordinate policies and investments to counter the malign use of these

technologies by authoritarian regimes; and provide concrete, competitive alternatives to

counter the adoption of digital infrastructure made in China.

• As part of the Emerging Technology Coalition, launch an International Digital

Democracy Initiative with allies and partners to align international assistance efforts

to develop, promote, and fund the adoption of AI and associated technologies that

comports with democratic values and ethical norms around openness, privacy, security,

and reliability.

• Implement a comprehensive U.S. national plan to support international technology

efforts around technical standards, foreign assistance, development finance, and export

controls.

• Enhance the United States’ position as an international emerging technology

research hub for collaborative R&D efforts by formalizing a partnership between

the U.S. National AI Research Institutes and multilateral initiatives like the Global

Partnership on AI (GPAI), creating a Multilateral AI Research Institute (MAIRI) in the

United States with key allies and partners, and catalyzing international collaboration and

talent exchanges.

Build an Emerging Technology Coalition. The United States should lead an Emerging

Recommendation

Technology Coalition (ETC) of like-minded nations either as part of a larger democracy

summit or as a stand-alone endeavor. The immediate step for the ETC should be to

organize its efforts to synchronize policies around the following seven critical areas:

244

CHAPTER 15

• Developing and operationalizing standards and norms, in support of democratic

values and the development of secure, reliable, and trusted technologies;

• Promoting and facilitating coordinated and joint R&D on AI and digital

infrastructure that advances shared interests and benefits humanity;

• Promoting democracy, human rights, and the rule of law through joint efforts to

counter censorship, malign information operations, human trafficking, and illiberal uses

of surveillance technologies;

• Exploring ways to facilitate data-sharing among allies and partners through enabling

agreements, common data archival procedures, cooperative investments in privacy-

enhancing technologies, and addressing legal and regulatory barriers;

• Promoting and protecting innovation, particularly through export controls, investment

screening, supply chain assurance, emerging technology investment, trade policy,

research and cyber protections, and intellectual property alignment;

• Developing AI-related talent, by analyzing labor market challenges, harmonizing

skills and certification requirements, and increasing talent exchanges, joint training, and

workforce development initiatives;

• Launching the International Digital Democracy Initiative.²

Launch an International Digital Democracy Initiative (IDDI). As part of the ETC, the United

Recommendation

States, with its allies and partners, should launch an IDDI to align international assistance

efforts to develop, promote, and fund the adoption of AI and associated technologies that

comport with democratic values and ethical norms around openness, privacy, security,

and reliability.

IDDI Strategy.

245

A FAVORABLE INTERNATIONAL TECHNOLOGY ORDER

The IDDI will be critical for enabling nations around the world to adopt secure, trusted, and

open digital ecosystems,³ empowering communities to use AI and digital technologies in

ways that strengthen democracies, promote sustainable development, and advance shared

values like privacy, human rights, and the rule of law. IDDI further provides an opportunity

for the United States and like-minded allies and partners to counter authoritarian uses

of AI, particularly by providing alternatives to digital infrastructure projects that are used

for illiberal ends, endanger the social cohesion among and between democracies, and

threaten collective security.⁴ As international digital and telecommunications infrastructure

investment needs continue to grow⁵ and China continues to use digital development to

export authoritarianism and expand influence, the United States and its allies and partners

must join forces to coordinate a strategy that maximizes the impact of government assistance

efforts and also catalyzes private-sector investment to address shared challenges.

“... the United States and its

allies and partners must join

forces to coordinate a strategy

that maximizes the impact of

government assistance efforts

and also catalyzes private sector

investment to address shared

challenges.”

Implement a comprehensive U.S. national plan to support international technology efforts.

Recommendation

The ISTS should include an integrated government-wide plan for using and bolstering

the tools of U.S. foreign policy—including technical and foreign assistance, development

financing, and export controls—to advance the ETC, the IDDI, and stand-alone projects.

As demonstrated below, the plan should include methods to shape international technical

standards; coordinate and expand programs of the Department of State, the United

States Agency for International Development, the U.S. International Development Finance

Corporation, and other federal agencies; and use targeted export controls to preserve key

U.S. and allied technical advantages and also further transparency and accountability. It

will require significant, dedicated appropriations to achieve meaningful results.⁶

246

A FAVORABLE INTERNATIONAL TECHNOLOGY ORDER

“The United States must maintain

its leadership in international

R&D by further establishing

itself as a hub of international

research into and involving

emerging technologies to foster

AI collaboration and coordination

with key allies and partners.”

overcome challenges to collaboration around cross-border data-sharing. Making the

United States an international emerging technology research hub has three components:

Components of

International Digital

Research Hub.

• First, the United States should provide formal research support to key international

efforts such as GPAI and the Organisation for Economic Co-operation and

Development,⁷ particularly through the National Science Foundation (NSF)’s National

AI Research Institutes.⁸ The important research undertaken by the National AI

Research Institutes—run by the NSF and other U.S. agencies—and by other United

States departments and agencies is an incredible resource that should support these

international efforts and advance AI and digital goals of the U.S. and like-minded

partners.

248

CHAPTER 15

• Second, the United States should work with key allies and partners to establish the

Multilateral AI Research Institute (MAIRI). MAIRI will facilitate joint efforts to develop

technologies that advance responsible, human-centric, and privacy-preserving AI/

machine learning (ML) that better societies and allow allies to pool their talents and

resources. It will provide a model for equitable, multilateral research, facilitate AI R&D

that builds on like-minded countries’ strengths, and foster a global AI workforce for the

next generation. MAIRI will be key to a U.S.-led effort to promote values of free and open

societies, win the global technology competition, unleash AI innovation and economic

prosperity, and develop AI applications that benefit humanity. MAIRI members will

champion agreed-upon research integrity principles, leverage trusted infrastructure

and research resources, and seek to be a part of a federated network of global research

institutes. NSF should be the anchor partner, but MAIRI should be structured to enable

participation of other federal agencies, like the Departments of State and Energy.⁹ The

United States should fund the initial startup costs, including acquisition of MAIRI’s

physical center located in the United States.

• Third, the United States should leverage existing O and J visa programs to facilitate

foreign researchers’ involvement in joint projects. Sustained, strong collaboration

between the United States and allies and partners is critical to winning this techno-

competition and unleashing innovation and entrepreneurship across like-minded

countries. There is no substitute for shoulder-to-shoulder research for building

relationships, exchanging ideas and expertise, and sparking future collaboration.¹⁰

Recommendation

Reorient U.S. foreign policy and the Department of State for great power competition in

the digital age. New outward-facing digital foreign policy initiatives are only part of the

equation for ensuring the long-term success of global technology policy. The United States

must make inward-focused reforms to the Department of State as well. There is currently

no clear lead for emerging technology policy or diplomacy within the State Department,

which hinders the Department’s ability to make strategic technology policy decisions. It

also creates confusion for allies and partners, who regularly express uncertainty regarding

which senior official should be their primary point of contact for issues related to key topics

such as AI, 5G, quantum computing, biotechnology, or new emerging technologies.

Competitive diplomacy in AI and emerging technology arenas is a strategic imperative

in an era of great power competition and necessitates an intensified reorientation of the

Department of State. The United States must redesign the internal structure, focus, and

culture of the State Department to adapt U.S. diplomacy for the digital age and empower

diplomats to advance American interests at the intersection of technology, security,

commerce, and human rights. Supporting these efforts and succeeding in U.S. diplomacy

will require targeted appropriations from Congress.

The Commission recommends the following immediate actions to reorient U.S. diplomacy:

• First, the Deputy Secretary of State for Management and Resources (D/MR)

should have responsibility to prioritize reorienting and reorganizing the Department

for technology diplomacy. Past administrations have used the D/MR position to lead

on strategic priorities and ensure execution. The D/MR should provide direction for

immediate and long-term planning around technology diplomacy, including policy

development, coordination, and resourcing. The D/MR should also have a lead role in

oversight and implementation of the ISTS.

249

A FAVORABLE INTERNATIONAL TECHNOLOGY ORDER

“The United States must redesign

the internal structure, focus, and

culture of the State Department

to adapt U.S. diplomacy for the

digital age ...”

•

Second, the State Department should expedite and prioritize efforts to staff,

resource, and build out the newly created Bureau of Cyberspace Security and

Emerging Technologies (CSET Bureau). The CSET Bureau, approved in early January

2021, would be led by an official with the title of Ambassador-at-Large and Coordinator.

The bureau would have a critical role as the focal point for U.S. diplomatic efforts

around security challenges associated with emerging technologies and would provide

an accountable home for AI advocacy within the Department.¹¹ The Department,

with congressional support, must ensure the CSET Bureau is adequately staffed and

resourced. Quickly standing up CSET is critical to building the Department’s technology

diplomacy capacity, improving technology policy coordination across the Department,

and regularly elevating technology issues to the attention of senior leaders. The

Department should assess where CSET should be placed to best achieve those

objectives but must ensure its creation is not further delayed.

•

Third, the State Department should enhance its presence in foreign and U.S.

technology hubs with a cadre of dedicated technology officers at U.S. missions to

strengthen diplomatic advocacy, improve technology scouting, and inform policy and

foreign assistance.

•

Fourth, AI-related technology modules should be incorporated into Foreign

Service Institute training courses at multiple levels to ensure U.S. diplomats are

equipped to lead in an environment being transformed by emerging technology.

•

Fifh, Congress must appropriate funds necessary for urgent State Department

needs both to augment the U.S. diplomatic corps and to support critical State

Department programs focused on AI and emerging technologies to advance U.S.

interests.

These steps are essential, but not sufficient, to further U.S. interests in tech diplomacy.

Eventually, the D/MR role should transition into a permanent Under Secretary for Science,

Research and Technology (State/Q). State/Q would lead a reorganization of the Department,

combining offices and bureaus to further a robust, coordinated approach to science and

technology diplomacy and foreign assistance in the context of great power competition.¹²

250

CHAPTER 15

Chapter 15 - Endnotes

¹ The threat to allied cohesion also extends to the military realm, insofar as building divergent or

incompatible digital systems poses challenges for interoperability or creates risks for U.S. forces

operating in allied countries. See Daniel Kliman, Why the United States Needs a Digital Development

Fund, Center for a New American Security at 2 (Oct. 10, 2019), https://www.cnas.org/publications/

commentary/why-the-united-states-needs-a-digital-development-fund (“Over the long term, China’s

digital investments could render some developing countries off-limits to U.S. forces, constricting the

geography of American military access.”).

² Detail on these critical areas can be found in the Chapter 15 Blueprint for Action andits associated

Annex.

³ USAID’s Digital Strategy defines the “digital ecosystem” as the “stakeholders, systems, and

enabling environments that together empower people and communities to use digital technology to

gain access to services, engage with each other, or pursue economic opportunities.” This includes

“a sound enabling environment and policy commitment; robust and resilient digital infrastructure;

capable digital service providers and workforce; and, ultimately, empowered end-users of digitally

enabled services.” Digital Strategy 2020-2024, USAID at 4 (June 2020), https://www.usaid.gov/usaid-

digital-strategy.

⁴ The Chinese government’s global infrastructure projects and its widespread state influence within

its private sector have enabled Chinese firms to provide surveillance and smart-city technologies to

hundreds of cities globally, particularly in developing countries, bolstering autocratic regimes and

enabling Chinese geopolitical coercion and government data collection. See, e.g., Hugh Harsono,

China’s Surveillance Technology is Keeping Tabs on Populations Around the World, The Diplomat

(June 18, 2020), https://thediplomat.com/2020/06/chinas-surveillance-technology-is-keeping-tabs-

on-populations-around-the-world/; Testimony of Steven Feldstein before the U.S.-China Economic and

Security Review Commission, Hearing on China’s Strategic Aims in Africa (May 8, 2020), https://www.

uscc.gov/sites/default/files/Feldstein_Testimony.pdf.

⁵ To support the G20, the Global Infrastructure Hub has forecasted global telecommunications

infrastructure investment needs at $8.9 trillion over the next approximately 20 years, with current

trends falling short of these needs by $1 trillion. Forecasting Infrastructure Investment Needs and

Gaps, Global Infrastructure Hub (last accessed Jan. 13, 2021), https://outlook.gihub.org/.

251

A FAVORABLE INTERNATIONAL TECHNOLOGY ORDER

⁶ Detailed recommendations for U.S. agencies and Congress can be found in the Chapter 15 Blueprint

for Action.

⁷ GPAI was launched in 2020 to “foster responsible development of AI grounded in these principles

of human rights, inclusion, diversity, innovation and economic growth.” Current members include

Australia, Brazil, Canada, the European Union, France, Germany, India, Italy, Japan, Mexico, the

Netherlands, New Zealand, Poland, Singapore, Slovenia, South Korea, Spain, the United Kingdom,

and the United States, with the OECD and UNESCO as Permanent Observers. GPAI bridges “the

gap between theory and practice,” particularly through research and technical expertise shared via

multi-stakeholder working groups. About GPAI, GPAI (last accessed Jan. 6, 2020), https://www.gpai.

ai/about/; UNESCO Joins Global Partnership on Artificial Intelligence as Observer, UNESCO (Dec. 10,

2020), https://en.unesco.org/news/unesco-joins-global-partnership-artificial-intelligence-observer.

⁸ Artificial Intelligence at NSF, NSF (Aug. 26, 2020), https://www.nsf.gov/cise/ai.jsp.

⁹ For example, the Department of Energy may provide critical expertise on undertaking applied

research with industry or through its national laboratories, particularly on high-performance and

quantum computing, while the Department of State can provide foreign policy expertise and support

initiatives on data-sharing and AI research clouds with allies and partners.

¹⁰ Detailed recommendations for each of these components can be found in the Chapter 15 Blueprint

for Action.

¹¹ Secretary Pompeo Approves New Cyberspace Security and Emerging Technologies Bureau, U.S.

Department of State (Jan. 7, 2021), https://2017-2021.state.gov/secretary-pompeo-approves-new-

cyberspace-security-and-emerging-technologies-bureau//index.html.

¹² These components of the State Department should include key functions of the CSET Bureau; the

Bureau of Oceans, Environment and Science; the Office of the Science and Technology Adviser; the

Coordinator for Cyber Issues; and the Center for Analytics.

252

CHAPTER 16

The Commission’s work ends where it began, with the conclusion

that artificial intelligence (AI) will transform virtually every aspect of

our existence. However, leadership in AI is not an end unto itself; it

is a necessary, but not sufficient, condition for the overarching goal

of preserving U.S. leadership in technology. That reality presents a

challenge for U.S. strategy: how to prioritize investments in AI and

other key emerging technologies and support specific projects

that will build on and amplify cross-technology strengths. The

United States must view its efforts to lead in AI through the broader

lens of competition across a range of emerging technologies, and,

therefore, also support a comprehensive strategy to sustain U.S.

leadership in key associated technologies.

“The United States must view

its efforts to lead in AI through

the broader lens of competition

across a range of emerging

technologies ...”

Leadership in AI relies on and drives leadership across a suite of emerging technologies.

AI sits at the center of the constellation of emerging technologies, enabling some and being

enabled by others.¹ For instance, 5G and quantum computing are poised to enable new

growth in AI capabilities, while AI stands to transform the biological sciences, producing

significant technological breakthroughs and turning the biotechnology sector into one of

the primary drivers of overall economic competitiveness.²

255

ASSOCIATED TECHNOLOGIES

AI is at the center of a constellation of emerging technologies, enabling some and

enabled by others. The United States Government must view its efforts to lead in AI

as part of a broader strategy to compete in each of these technologies:

The United States

Government must:

Define and prioritize the key emerging

Associated

technologies needed to ensure U.S.

Technologies.

national competitiveness, including U.S.

leadership in the following technologies

and platforms:

• Advanced biofabrication capabilities

• Quantum chip fabrication

•

5G spectrum sharing

• Robotics software

• Additive manufacturing

• Energy storage technologies

China is pursuing a comprehensive technology leadership strategy. China’s strategic

investments in key sectors through its Made in China

2025 initiative threaten U.S.

technological prowess, economic prosperity, and national security.³ In addition to

investments in AI, China is seeking to become a world leader in quantum, 5G, and

biotech, among other areas, and sees its strategies to lead in AI and each of these other

technologies as mutually reinforcing. It has made clear which technologies it views as key

national priorities and is investing heavily in a wide range of sectors that it assesses are

essential to overall technical leadership.

The United States has neither identified, nor prioritized leadership in, the technologies that

are central to national competitiveness. The United States must develop a unified list of

the technologies that will underpin national competitiveness in the 21st century because

the first-mover advantage of developing and deploying technologies like microelectronics,

biotechnology, and quantum computing will make it difficult for the United States to catch

up to China. The lack of such a list results in disparate funding and policy approaches to

technology protection and promotion across the U.S. government. The absence of clear

priorities also makes it more difficult to effectively marshal private-sector investment in

key technologies. In critical sectors with strong network effects like telecommunications,

a winner-take-all dynamic raises the stakes even further for rapidly developing a leading

technology platform.⁴

256

CHAPTER 16

Ensuring U.S. leadership in the manufacturing of key emerging technology platforms

will be an essential component of national competitiveness. Identifying and supporting

research in priority technologies is a necessary but insufficient step to maintain national

competitiveness in emerging technologies. The United States will also have to invest in the

production of strategic physical elements of these technologies to create game-changing

platforms, maximize U.S. competitiveness, and reduce dependencies that create national

security vulnerabilities. Such investments are often expensive, but a strategic approach

does not require manufacturing every advanced component domestically and will pay

tremendous long-term dividends. The need to support advanced manufacturing applies

to nearly every key emerging technology sector, including semiconductors, quantum

computing, biotechnology, telecommunications equipment, and others, and is reflected in

the following recommendations.

Technology leadership will require major new investments in underlying digital infrastructure.

It is impossible to ignore the state of America’s underlying digital infrastructure when

considering a strategy to preserve overall U.S. leadership in technology. The sophistication

and reach of the core U.S. digital infrastructure that underpins connectivity, namely

high-speed internet and telecommunications networks, significantly lag behind those of

many other developed nations.⁵ Additionally, U.S. physical infrastructure remains largely

disconnected; no U.S. cities are ranked among the top 10 in smart-city connectedness,

and only one is in the top 30.⁶ Maximizing citizens’ access to the digital economy, ensuring

they have the requisite digital skills, and more closely connecting the physical and digital

worlds will be necessary to fuel future growth. Boosting the digital connectivity of physical

U.S. assets will not only enhance their effectiveness and reliability, but also generate new

sources of data that enable novel, potentially revolutionary uses of AI in areas ranging

from energy grid management and urban planning to transportation. As the United

States considers options to modernize U.S. physical infrastructure, prioritizing its digital

connectivity will provide substantial long-term benefits and buttress U.S. technology

competitiveness and national security.

Identifying and Prioritizing Technologies Central to National Competitiveness.

While the United States should by no means adopt China’s centrally planned and state-

directed economic model, it must start by developing better strategic planning, forecasting,

and prioritization of emerging technologies to ensure long-term competitiveness. The

government should:

Define and prioritize the key emerging technologies that are needed to ensure U.S. national

Recommendation

competitiveness. As part of its National Technology Strategy (see Chapter 9 of this report),

the White House should publish a list of critical and emerging technologies in which U.S.

leadership is essential. It should develop detailed implementation plans for each sector to

determine how the government should best work with industry to promote U.S. leadership,

assess which specific subsectors are crucial to national security, and determine what

257

ASSOCIATED TECHNOLOGIES

“As part of its National

Technology Strategy (see

Chapter 9), the White House

should publish a list of critical

and emerging technologies

in which U.S. leadership

is essential.”

regulatory steps or incentives are necessary to create the required investment environment.

These plans should promote investment in specific platforms that will have a force multiplier

effect on U.S. technology leadership, identify key choke points where competitors could

potentially be blocked with minimal impact on U.S. industry, and promote supply chain

resiliency. The creation and maintenance of such a list and the associated implementation

plans will help produce a national consensus across government, industry, and academia

about which sectors are most important in the emerging techno-economic competition.

The result will be an important message to Congress regarding where the country must

prioritize and expend resources, as well as a powerful demand signal to industry.

Many similar lists exist throughout the government, but there has been no effort to unify

them into a single, authoritative document accompanied by a strategic vision and detailed

follow-through actions designed to ensure long-term U.S. leadership.⁷ However, the

significant overlap between these lists demonstrates an emerging national consensus on

which technologies are most critical to U.S. national competitiveness. Table 5 illustrates

the initial slate of technologies that the Commission recommends including as part of a

broader technology leadership strategy, as well as whether or not those technologies have

been included on select, existing U.S. government lists of critical technologies. As an initial

step, the Commission recommends that the White House designate these technologies as

critical through an Executive Order and direct Departments and Agencies to prioritize and

coordinate them accordingly.

258

CHAPTER 16

U.S. Government Lists of Critical Technologies

NSCAI-

2018

DoD List

Commerce

PCAST List

S.3832 -

WH Nat

Proposed

National

of Critical

ANPRM on

of Industries

Endless

Strategy

Critical

Defense

Emerging

of the Future

Frontier

for

Technology List

Strategy

Technologies

Act

C&ET

Artificial

Intelligence

Biotechnology

Quantum

Computing

Semiconductors

and Advanced

Hardware

Autonomy and

Robotics

5G and

Advanced

Networking

Advanced

Manufacturing

Energy

Systems

Actions to Promote Technologies and Platforms Essential

to U.S. Leadership and National Security.

After reaching consensus on the set of emerging technologies essential to overall U.S.

technology leadership, the Executive Branch should assess each sector and identify

specific platforms that meet the following criteria:

• Have potential applications of strategic and national security importance;

• Could have a significant impact on overall U.S. technical leadership and

competitiveness, either alone or when combined with existing U.S. technical strengths;

and

• Require government action to spur or protect its development.

Such platforms could require government support for several reasons. In some instances,

a market failure may lead to underinvestment by the private sector in an area of strategic

importance to national security. In other instances, seizing a market opportunity may only

be possible if the federal government focuses the private sector, academia, and research

organizations on a specific goal. The government must tailor its approach to the context

259

ASSOCIATED TECHNOLOGIES

by increasing funding, implementing regulatory changes, or taking other steps aimed at

promoting innovation and protecting advantages that fit the circumstances.

The Commission has already presented recommendations to support U.S. leadership

in key technology platforms within several of the aforementioned strategic technologies.

For example, Chapter 11 of this report recommends establishing a National AI Research

Resource, which would create an essential platform to sustain and extend U.S. leadership

in AI. Additionally, in Chapter 13 of this report, the Commission provided a series of

recommendations for promoting U.S. leadership in microelectronics, including specific

actions to incentivize the construction of a leading-edge merchant fabrication facility

domestically.

The recommendations below build on the Commission’s previous work and provide

further actions the U.S. government could take to promote U.S. leadership in the key

associated technologies and platforms that the Commission assesses to be of greatest

strategic importance—specifically, biotechnology, quantum computing, 5G and advanced

networking, autonomy and robotics, advanced and additive manufacturing, and energy

systems.⁸

Biotechnology.

Biology is now programmable, and AI’s ability to identify ways to optimize this programming

will enable transformational biotechnology breakthroughs. AI was crucial in the rapid

development of COVID-19 vaccines, allowing researchers to finalize the genetic sequence

of a vaccine candidate only two days after the virus’ full genetic sequence was first posted

online.⁹ Computer vision techniques applied to medical imagery have also enabled more

accurate and efficient diagnoses.¹⁰ And recently, an AI network made substantial progress

over the last year toward solving one of biology’s most daunting challenges: determining

a protein’s 3D shape from its amino-acid sequence.¹¹ Tools such as these will become

“Biology is now programmable,

and AI’s ability to identify ways

to optimize this programming

will enable transformational

biotechnology breakthroughs.”

260

CHAPTER 16

even more powerful in combination with synthetic biology and gene editing. Together they

will enhance human health by allowing deeper studies of the building blocks of life and

enabling the quicker discovery and fabrication of more advanced drugs and materials. As

AI fuels rapid new developments in the biological sciences and biotechnology becomes

a greater driver of the overall world economy, the strategic consequences of ceding

leadership in biotechnology will increase significantly—a fact that the COVID-19 pandemic

illustrates in clear and stark terms. The government should:

Recommendation

Prioritize the development of an advanced domestic biotechnology R&D ecosystem. As

part of a national bioeconomy strategy, the United States should support the development

of biotechnology platforms that maximize researchers’ ability to utilize AI to drive new

biological breakthroughs and help transition advanced research into physical products

at scale. This will necessitate support for both world-class biodata resources to fully

harness the power of AI and biomanufacturing platforms to rapidly realize the benefits

from analytical breakthroughs:

•

Biodata: The United States should fund and prioritize efforts to build a world-class

biobank containing a wide range of high-quality biological and genetic data sets

securely accessible by researchers. GenBank, the leading U.S. genetic database, which

is run by the National Institutes of Health, is currently underfunded, underutilized, and

poorly curated. The goal should be to create a genetic database that is well-curated and

easy for researchers to access and use; contains a significant number and broad range

of whole human, animal, and plant genomes; and aggregates open and proprietary

data sets across government and the private sector. It should also contain de-identified

metadata about corresponding phenotypes whenever possible and include strong

privacy protections for human genetic data. This will require significant levels of

funding; China National GeneBank, the equivalent facility in China that is operated

by the BGI Group (formerly the Beijing Genomics Institute), required approximately

$117 million in initial funding.¹² Establishing such an entity in the United States would

enhance and democratize biotechnology innovations by pooling existing data resources

and facilitating new levels of AI-enabled analysis of genetic data while also reducing

U.S. researchers’ reliance on BGI or other Chinese entities for access to large-scale

genomic databases for research.

•

Biomanufacturing: The United States should support efforts to diversify and expand

the biotechnology industry beyond its current vertically integrated models and

encourage the development of multiple standardized, merchant biofabrication

facilities. Doing so is necessary to ensure U.S. biomanufacturing capabilities keep pace

with AI’s transformative impact on the bioeconomy. Expanding access to advanced

biofabrication tools among startups and laboratories would allow firms to rapidly design

new molecules and materials via the cloud and place immediate orders for fabrication.

Such efforts should include R&D funding and incentives to support advanced biotech

manufacturing initiatives through entities such as the Biomedical Advanced Research

and Development Authority (BARDA),¹³ with appropriate stewardship, and the expansion

of existing relevant programs such as BioMADE.¹⁴ Congress should prioritize such

initiatives in future health-related spending bills. Given that up to 60% of the physical

inputs to the global economy could be produced via synthetic biology, there is a clear

and pressing need for the United States to retain leadership in biomanufacturing moving

forward.¹⁵

261

ASSOCIATED TECHNOLOGIES

“... quantum computers have

the potential to outperform

their classical counterparts on

certain classes of problems

related to machine learning and

optimization, the simulation

of physical systems, and the

collection and transfer of

sensitive information.”

Quantum Computing.

As the pace of innovation predicted by Moore’s Law becomes increasingly difficult for

semiconductor manufacturers to maintain due to the physical limits of microchip design,

leadership in next-generation computer hardware will be essential to preserving long-term

U.S. advantages in strategic technologies like AI.¹⁶ Although classical computers will likely

remain the most economical way of performing day-to-day computational tasks in the near

future, quantum computers have the potential to outperform their classical counterparts

on certain classes of problems related to machine learning (ML) and optimization, the

simulation of physical systems, and the collection and transfer of sensitive information.

For example, quantum computers may be able to efficiently optimize military logistics or

discover new materials for weapon systems.¹⁷ Each of these applications creates novel

national security threats and opportunities at the intersection of AI and quantum computing.

The government should:

Transition from basic research to national security applications of quantum computing and

Recommendation

incentivize domestic fabrication. The United States is a global leader in research of quantum

computers, but it risks losing its edge in applications to national security. Recognizing

that advances in quantum computing may drive advances in AI, the United States must

establish trusted sources of materials and components for quantum computers, invest in

the development of hybrid quantum-classical algorithms, and focus on fielding of national

security applications. Offering access to both classical and quantum computers through

262

CHAPTER 16

the National AI Research Resource will facilitate the development of hybrid quantum-

classical algorithms that leverage noisy intermediate-scale quantum computers. Publicly

announcing specific government use cases of quantum computers will signal that a market

exists for national security applications and encourage further investment by the private

sector.

5G and Advanced Networking.

5G networks will form the connective tissue between AI platforms, which means maintaining

access to trusted and robust 5G networks is a critical component of overall leadership in

AI. Huawei is pursuing global dominance in 5G, and there is no single supplier that can

compete with it in terms of both price and quality. Due to the urgency of the issue, the

United States should pursue several complementary approaches concurrently to ramp up

deployment of 5G domestically and provide a credible alternative to Huawei. As a starting

point, any comprehensive effort should include support for dynamic spectrum sharing.¹⁸

The government should:

“Expanding spectrum sharing

efforts is critical to ensuring

that DoD maintains access

to spectrum essential for

operational effectiveness while

broadening commercial access to

spectrum for 5G networks.”

Bolster and accelerate U.S.

5G network deployment through mid-band spectrum

Recommendation

sharing. Expanding spectrum-sharing efforts is critical to ensuring that the Department

of Defense (DoD) maintains access to spectrum essential for operational effectiveness

while broadening commercial access to spectrum for 5G networks. A multi-agency effort

is needed to expand sharing arrangements and licenses and permit additional portions

of the mid-band to be simultaneously utilized by DoD and commercial carriers. Through

this portfolio approach, the United States stands the best chance of accelerating its 5G

deployment at a pace that can support the widespread adoption of AI.

263

ASSOCIATED TECHNOLOGIES

Autonomy and Robotics.

Autonomous systems are already unlocking value across global markets. In the private

sector, they enable products ranging from expert advisory systems and self-driving

vehicles to manufacturing. In the realm of national security, autonomous systems generate

opportunities to reduce the number of warfighters in harm’s way, increase the pace and

quality of decisions, and create entirely new military capabilities.¹⁹ The ability to design

and produce the hardware and software for advanced robotics is an essential part of

autonomous systems. The government should:

Incentivize the development of world-class software platforms for robotic and autonomous

Recommendation

systems. The future of autonomy and robotics will manifest in almost unlimited shapes and

sizes as firms develop and tailor robots for different use cases and environments. The

U.S. trails nations such as China, Japan, and South Korea in the deployment of robots

and robotic hardware and must work to improve its capabilities in such areas as materials

design and energy storage for robots.²⁰ However, U.S. expertise in software development

lends itself to creating a world-class digital platform for many classes of robotic hardware.

The software powering robotic systems will be built upon several core capabilities rooted

in AI: It will need to be able to sense its environment, reason, and operate in the world

around it.²¹ In creating cutting-edge software for these types of capabilities, there is an

opportunity for U.S. firms to win the market for the software platforms that power the next

wave of industrialization.²² To promote U.S. leadership in the development of software

for autonomous systems, the U.S. government should fuel industry’s ongoing efforts

by supplementing the basic R&D, standard-setting, and data-sharing programs led by

National Institute of Standards and Technology (NIST)’s Intelligent Systems Division.²³ It

should also incentivize early adoption of automation and create markets for autonomous

systems in areas already ripe for them, such as mail sorting, that will yield data and

experience relevant for achieving scale and addressing adjacent markets.²⁴ Combined,

a multi-pronged approach along these lines would position industry to compete more

effectively in the market for autonomous system software, a strategically important area

aligned with existing U.S. technical strengths.

Advanced and Additive Manufacturing.

The capacity to produce high-tech goods domestically is critical to national security, both

to maintain access to finished goods and as a driver of innovation. In terms of access,

the United States must strive for self-reliance in industries that are critical to national

security or that would take too long to regenerate in the event of protracted conflict.²⁵

Innovation also benefits from a tight feedback loop between technological design and

production, which allows for more rapid iteration.²⁶ This link is particularly important in

the defense sector, where feedback from the manufacturing process back into the R&D

cycle helps bring technology from lab to military operations. Longer-term disruptions to

the manufacturing industry through new techniques such as additive manufacturing also

264

CHAPTER 16

“The capacity to produce

high-tech goods domestically

is critical to national security,

both to maintain access to

finished goods and as a driver

of innovation.”

pose threats and opportunities for national security. For example, additive manufacturing

may enable a step-change in domestic manufacturing capabilities, but it also creates

new threats by potentially democratizing the production of firearms and other goods with

military applications.²⁷ The government should:

Accelerate additive manufacturing production of legacy parts across the DoD. Additive

Recommendation

manufacturing and 3D printing have the potential to transform manufacturing. They are

capable of rapid, high-quality, and complex production, and they are flexible enough that

3D printers may be able to be located near the point of need for just-in-time production.²⁸

Although current additive manufacturing techniques struggle to replicate the quality of

advanced traditional manufacturing techniques, AI has already shown the ability to enable

significant improvements in their accuracy.²⁹ The Federal Government should proactively

support initiatives that advance the development of additive manufacturing techniques

and also provide practical benefits by easing the production of legacy items.³⁰ The DoD

should announce a goal of identifying all legacy parts in active weapon systems that are

capable of being produced via additive manufacturing and 3D printing and doing so by

2025.

Energy Systems.

Cheap and reliable access to energy is critical to U.S. national security, whether it be to

ensure military readiness, facilitate the response to a domestic crisis, or keep the economy

functioning smoothly. As an input to nearly every sector, the price of energy directly impacts

economic output and is a key determinant of U.S. national competitiveness. Furthermore,

dependence on foreign countries for energy resources and technologies would put the

United States in a position of vulnerability, especially if those resources or technologies

are controlled by strategic competitors. Although the United States is at the forefront of the

exploration, extraction, and processing of oil and gas and possesses significant domestic

265

ASSOCIATED TECHNOLOGIES

reserves, China is far and away the leading producer of renewable energy and is investing

heavily in advanced energy storage technologies, such as batteries and their constituent

materials.³¹ To remain competitive, in these critical sectors U.S. industry will need to achieve

aggressive cost targets in terms of kilowatts/hour and energy density. This is especially

true in markets with the most substantial growth potential, such as long-duration stationary

storage devices and battery packs for electric vehicles.³² The government should:

Develop and domestically manufacture energy storage technologies to meet U.S. market

Recommendation

demand by 2030. Developing new technologies to more effectively store electrical

energy so it is readily available whenever and wherever needed would drive advances in

electricity transmission and distribution. It would also offer advantages to the United States

both economically and strategically. To accelerate breakthroughs in energy storage,³³

the Department of Energy has set the ambitious goal of developing and domestically

manufacturing storage technologies capable of meeting the entirety of U.S. market

demand by 2030.³⁴ Congress should fully fund the federal R&D and establish incentives

for commercialization needed to achieve the Department of Energy’s Energy Storage

Grand Challenge roadmap by 2030.³⁵

“Cheap and reliable access

to energy is critical to U.S.

national security, whether it be

to ensure military readiness,

facilitate the response to a

domestic crisis, or keep the

economy functioning smoothly.”

266

CHAPTER 16

Chapter 16 - Endnotes

¹ Recognizing this connection, Congress included AI and “associated technologies” as they relate to

national security within the scope of the Commission’s mandate.

² The Commission’s first Interim Report identified biotechnology, quantum computing, and 5G as key

emerging technologies associated with AI. See Interim Report, NSCAI at 50 (Nov. 2019), https://www.

nscai.gov/previous-reports/.

³ Made in China 2025 includes the following sectors: new-generation information technology,

high-grade machine tooling and robotics, aviation and aerospace equipment, marine engineering

equipment and high-tech ships, advanced rail transportation equipment, new energy automobiles,

electric power equipment, agriculture equipment, new materials, and biomedicine and high-tech

medical devices. See Alice Tse & Julianna Wu, Why ‘Made in China 2025’ Triggered the Wrath of

President Trump, South China Morning Post (Sept. 11, 2018), https://multimedia.scmp.com/news/

china/article/made-in-China-2025/index.html.

⁴ According to McKinsey, “Platforms are the back-end technology capabilities, whether provided by

individual systems or by assemblies of multiple systems, that power products.” Ross Frazier, et al.,

Products and Platforms: Is Your Technology Operating Model Ready?, McKinsey Digital (Feb. 28,

2020), https://www.mckinsey.com/business-functions/mckinsey-digital/our-insights/products-and-

platforms-is-your-technology-operating-model-ready.

⁵ The United States ranks 18th among 37 Organisation for Economic Co-operation and Development

(OECD) countries in fixed and mobile broadband subscriptions per 100 inhabitants and eighth

in average broadband speed. Broadband Portal, OECD (July 2020), https://www.oecd.org/sti/

broadband/broadband-statistics/ (see “Penetration and data usage” table “1.2 Fixed and mobile

broadband subscriptions per 100 inhabitants” [Dec. 2019] and “Speeds” table “5.2 Akamai average

speed” [Q1 2017]).

⁶ Smart City Index, IMD, 8 (Oct. 2019), https://www.imd.org/research-knowledge/reports/imd-smart-

city-index-2019/#:~:text=The%20Top%2010%20smartest%20cities,and%20Dusseldorf%20(10th).

⁷ For example, the White House published its National Strategy for Critical and Emerging Technologies

in October 2020, which included a list of critical and emerging technologies as identified by

Departments and Agencies. The document does not provide detail on how each technology is

essential to national competitiveness and lacks a specific plan for promoting and protecting U.S.

advantages in each. See National Strategy for Critical and Emerging Technologies, The White House

at A-1 (Oct. 2020), https://www.whitehouse.gov/wp-content/uploads/2020/10/National-Strategy-for-

CET.pdf?utm_source=morning_brew. In their 2018 report, Michael Brown and Pavneet Singh argue

that the lack of a unified list of critical technologies harms the ability of the United States to protect

against technology transfer. See Michael Brown & Pavneet Singh, China’s Technology Transfer

Strategy, Defense Innovation Unit Experimental at 37 (Jan. 2018), https://admin.govexec.com/media/

diux_chinatechnologytransferstudy_jan_2018_(1).pdf.

⁸ This includes all technologies other than AI on NSCAI’s proposed critical technologies list with the

exception of semiconductors, which are addressed separately in Chapter 13 of this report. Elements

of AI-enabled biotechnology are also separately addressed in Chapter 1 of this report. Additional

recommendations to promote U.S. leadership in biotechnology, quantum computing, and 5G can be

found in the Chapter 16 Blueprint for Action.

⁹ Hannah Mayer, et al., AI Puts Moderna Within Striking Distance of Beating COVID-19, Harvard

Business School (Nov. 24, 2020), https://digital.hbs.edu/artificial-intelligence-machine-learning/ai-

puts-moderna-within-striking-distance-of-beating-covid-19/; Noah Weiland, et al., Modern Vaccine Is

Highly Protective Against Covid-19, the F.D.A. Finds, New York Times (Dec. 18, 2020), https://www.

nytimes.com/2020/12/15/health/covid-moderna-vaccine.html.

¹⁰ Junfeng Gao, et al., Computer Vision in Healthcare Applications, Journal of Healthcare Engineering

(March 4, 2018), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5857319/.

¹¹ Ewen Callaway, ‘It will Change Everything’: DeepMind’s AI Makes Gigantic Leap in Solving Protein

Structures, Nature (Nov. 30, 2020), https://www.nature.com/articles/d41586-020-03348-4.

¹² Zhuang Pinghui, China Opens First National Gene Bank, Aiming to House Hundreds of Millions

of Samples, South China Morning Post (Sept. 22, 2016), https://www.scmp.com/news/china/

article/2021623/chinas-noahs-ark-first-national-gene-bank-opens-shenzhen.

267

ASSOCIATED TECHNOLOGIES

¹³ Biomedical Advanced Research and Development Authority (BARDA), U.S. Department of Health

and Human Services (Dec. 4, 2019), https://www.hhs.gov/about/agencies/orgchart/aspr/barda/index.

html.

¹⁴ BioMADE is a public-private partnership with DoD, operated under Manufacturing USA, that

is focused on building a sustainable bioindustrial manufacturing ecosystem and enhancing U.S.

bioindustrial competitiveness. See BioMADE (Bioindustrial Manufacturing and Design Ecosystem),

Manufacturing USA (last accessed Jan. 9, 2021), https://www.manufacturingusa.com/institutes/

biomade.

¹⁵ Michael Chui, et al., The Bio Revolution, McKinsey Global Institute at 43 (May 13, 2020), https://

www.mckinsey.com/industries/pharmaceuticals-and-medical-products/our-insights/the-bio-revolution-

innovations-transforming-economies-societies-and-our-lives.

¹⁶ Steve Blank, What the GlobalFoundries’ Retreat Really Means, IEEE Spectrum (Sept. 10, 2018),

https://spectrum.ieee.org/nanoclast/semiconductors/devices/what-globalfoundries-retreat-really-

means.

¹⁷ Pontus Vikstål, et al., Applying the Quantum Approximate Optimization Algorithm to the Tail-

Assignment Problem, Physical Review Applied, vol. 14, iss. 3 (Sept. 3, 2020), https://doi.org/10.1103/

PhysRevApplied.14.034009; He Ma, et al., Quantum Simulations of Materials on Near-term Quantum

Computers, npj Computational Materials (July 2, 2020), https://doi.org/10.1038/s41524-020-00353-z.

¹⁸ The 5G Ecosystem: Risks & Opportunities for DoD, DoD Defense Innovation Board (April 2019),

https://media.defense.gov/2019/Apr/03/2002109302/-1/-1/0/DIB_5G_study_04.03.19.pdf.

¹⁹ Summer Study on Autonomy, DoD Defense Science Board (June 2016), https://dsb.cto.mil/

reports/2010s/DSBSS15.pdf.

²⁰ Maximiliano Dvorkin & Asha Bharadwaj, Which Countries and Industries Use the Most Robots,

Federal Reserve Bank of St. Louis (Nov. 7, 2019), https://www.stlouisfed.org/on-the-economy/2019/

november/robots-affecting-local-labor-markets.

²¹ Advanced materials (such as biological components), brain-computer interfaces, and small and

efficient power supplies are additional areas of potential innovation that connect robotics to AI and

other associated technologies described in this chapter.

²² For example, core capabilities might include gripping physical objects, which robotics maker ABB

and several firms in the U.S. and Europe are currently pursuing. See Jonathan Vanian, Industrial

Robotics Giant Teams Up with a Rising A.I. Startup, Fortune (Feb. 25, 2020), https://fortune.

com/2020/02/25/industrial-robotics-ai-covariant/.

²³ Intelligent Systems Division, NIST (last accessed Jan. 6, 2021), https://www.nist.gov/el/intelligent-

systems-division-73500.

²⁴ One specific area to expand demand for autonomous systems could be drastically scaling the

U.S. Postal Service’s Autonomous Mobile Robot pilot program from 25 sorting facilities to all sorting

facilities by 2025. Autonomous Mobile Robots and the Postal Service, USPS Office of Inspector

General (April 9, 2018), https://www.uspsoig.gov/sites/default/files/document-library-files/2019/RARC-

WP-18-006.pdf.

²⁵ Critical Technology Accessibility, National Academies Press (2006), https://www.nap.edu/

read/11658/chapter/1; see also Assessing and Strengthening the Manufacturing and Defense

Industrial Base and Supply Chain Resiliency of the United States, Interagency Taskforce in Fulfillment

of Executive Order 13806 at 46 (Sept. 2018), https://media.defense.gov/2018/oct/05/2002048904/-

1/-1/1/assessing-and-strengthening-the-manufacturing-and%20defense-industrial-base-and-supply-

chain-resiliency.pdf (identifying 10 risk archetypes threatening America’s manufacturing and defense

industrial base).

²⁶ Strategy for American Leadership in Advanced Manufacturing, National Science & Technology

Council (Oct. 2018), https://trumpwhitehouse.archives.gov/wp-content/uploads/2018/10/Advanced-

Manufacturing-Strategic-Plan-2018.pdf; Gregory Tassey, Rationales and Mechanisms for Revitalizing

US Manufacturing R&D Strategies, NIST (Jan. 29, 2010), https://www.nist.gov/system/files/

documents/2017/05/09/manufacturing_strategy_paper_0.pdf.

268

CHAPTER 16

Chapter 16 - Endnotes

²⁷ 3D Opportunity for Adversaries, Deloitte (Aug. 22, 2017), https://www2.deloitte.com/us/en/insights/

focus/3d-opportunity/national-security-implications-of-additive-manufacturing.html.

²⁸ Audit of DoD’s Use of Additive Manufacturing for Sustainment Parts, DoD Inspector General (Oct.

17, 2019), https://media.defense.gov/2019/Oct/21/2002197659/-1/-1/1/DODIG-2020 - 003.pdf.

²⁹ Mark Anderson, 3D Print Jobs Are More Accurate with Machine Learning, IEEE Spectrum (Feb. 19,

2020), https://spectrum.ieee.org/tech-talk/artificial-intelligence/machine-learning/3d-print-jobs-news-

accurate-machine-learning.

³⁰ For instance, in August 2020, the DoD printed the first metal part for a B-52 jet engine. Kyle

Mizokami, The Old School Engine That Powers the B-52 Gets a 3D-Printed Upgrade, Popular

Mechanics (Aug. 10, 2020), https://www.popularmechanics.com/military/aviation/a33535790/air-

force-3d-print-metal-part-turbofan-engine/.

269

ASSOCIATED TECHNOLOGIES

³¹ Robert Rapier, Ten Countries That Dominate World Fossil Fuel Production, Forbes (July 14, 2019),

https://www.forbes.com/sites/rrapier/2019/07/14/ten-countries-that-dominate-fossil-fuel-production;

Country Rankings, International Renewable Energy Agency (last accessed Jan. 6, 2021), https://www.

irena.org/Statistics/View-Data-by-Topic/Capacity-and-Generation/Country-Rankings.

³² Energy Storage, U.S. Department of Energy (last accessed Jan. 6, 2021), https://www.energy.gov/

oe/energy-storage.

³³ The field of energy storage includes a broad technology base such as batteries (both conventional

and advanced), electrochemical capacitors, flywheels, power electronics, control systems, and

software tools for storage optimization and sizing.

³⁴ Energy Storage, U.S. Department of Energy (last accessed Jan. 6, 2021), https://www.energy.gov/

oe/energy-storage.

³⁵ Energy Storage Grand Challenge: Roadmap, U.S. Department of Energy (Dec. 2020), https://www.

energy.gov/sites/prod/files/2020/12/f81/Energy%20Storage%20Grand%20Challenge%20Roadmap.

pdf.

270

THE NATIONAL SECURITY COMMISSION ON ARTIFICIAL INTELLIGENCE

Chapter 1: Emerging Threats in the AI Era

273

Combating Malign Information Operations Enabled by AI

273

reparing for AI-Enabled Cyber Conflict

278

Chapter 2: Foundations of Future Defense

291

Chapter 3: AI and Warfare

337

Chapter 5: AI and the Future of National Intelligence

349

Chapter 6: Technical Talent in Government

357

Chapter 7: Establishing Justified Confidence in AI Systems

379

Chapter 8: Upholding Democratic Values: Privacy, Civil

395

Liberties, and Civil Rights in Uses of AI for National Security

The following Blueprints for Action cover Part I of NSCAI’s Final Report. Part I, “Defending America in the

AI Era,” (Chapters 1-8) outlines what the United States must do to defend against the spectrum of AI-

related threats from state and non-state actors, and recommends how the U.S. government can

responsibly use AI technologies to protect the American people and our interests. These Blueprints for

Action complement the Commission’s Final Report and mirror its organizational structure.

The Blueprints for Action serve as more detailed roadmaps for Executive and Legislative branch actions

to retain America’s AI leadership position. They identify who should take a particular action--Congress,

the White House, or an executive branch department or agency. The Commission provides estimated

increases in funding or appropriations as part of its recommendations. All recommendations that

include funding figures should be considered estimates for consideration by Congress and/or the

Executive branch.

This report does not contain a separate Blueprint for Action for Chapter 4. This is because given the importance of the

topic, the Commission chose to detail its arguments, recommendations, and the specific actions required to implement

them directly in this chapter. Additionally, further detail on how the United States should adapt its TEVV policies to maintain

confidence in AI systems can be found in Chapter 7 and its associated Blueprint for Action, and recommendations on

relevant changes to DoD organizational structure can be found in Chapter 3.

272

BLUEPRINT FOR ACTION: CHAPTER 1

Chapter 1: Emerging

Threats in the AI Era

Blueprint for Action: Number One

Combating Malign Information Operations Enabled by AI.

The use of AI to produce, manipulate, and promote malign information marks a disruptive

evolution in the use of information as a tool of statecraft, a weapon of war, and a threat to

democracy.¹ The following recommendations represent a strategic, organizational, and

operational framework that the U.S. government should adopt to adequately defend and

counter malign information operations in the AI era, including by employing AI-enabled

technologies.

Recommendation: A National Strategy for the Global Information Domain

Recommendation

Expanding upon the principles of information statecraft outlined in the 2017 National

Security Strategy,² the President should issue a new national strategy for the global

information domain that more fulsomely addresses how AI and associated technologies

are defining new fronts in this area. The strategy should:

• Acknowledge that the network-connected world is dissolving barriers between

societies.

• Prioritize the global information domain as an arena for competition.

• Detail how adversarial state and non-state actors are attempting to define and control

the global information domain in order to shape global opinion and achieve strategic

advantage.

• Account for the critical role of AI-enabled malign information in achieving these goals.

• Designate malign information operations as a national security threat with its own set of

priority actions to defend, counter, and compete against them.

• As necessary, update critical infrastructure designations and require relevant

departments and agencies to update sector-specific plans to reflect emerging

technologies.

• Establish organizational structures for U.S. national security agencies to defend, counter,

and compete against the threat.

Action for the President:

• Issue a supplemental National Strategy for the Global Information Domain.

273

EMERGING THREATS IN THE AI ERA

Action for Congress:

• Congress should direct the Executive Branch to transmit a National Strategy for the

Global Information Domain that categorizes the global information domain as an arena

of competition vital to the national security of the United States.

Organizational Framework

The proliferation of malign information has exposed an Achilles heel in the U.S. national

security apparatus. Previous major reorganizations could not foresee contemporary digital

technology and society’s profound dependence upon it. They could not anticipate the use

of ICT platforms and tools, bots, and AI-enabled technologies to spread false information.

They do not account for the role that the commercial sector and civil society play in

defending against malign information, and enabling its spread. Individual agencies such

as the Department of Homeland Security (DHS), the Federal Bureau of Investigation (FBI),

and the Office of the Director of National Intelligence (ODNI) have stretched their mandates

to confront the threat. They rely on narrow sets of outdated tools, and are hampered by

cultures shaped by the Cold War and counter-terrorism paradigms.

Recommendation: Create a Joint Interagency Task Force (JIATF) and Operations Center.

Recommendation

Action for the President:

• Direct creation of a JIATF and operations center to lead and integrate government

efforts to counter foreign-sourced malign information in real time.

o The Presidential action should direct the Secretaries of State, Defense, Justice, and

Homeland Security, the Attorney General and the Director of National Intelligence,

to create a JIATF and stand-up an operations center to counter foreign-sourced

malign information.

o The JIATF should integrate efforts of key offices, bureaus, and divisions within each

of these agencies, as well as the broader intelligence community (IC) and law

enforcement establishment.

o The JIATF should have the responsibility to survey the landscape of relevant public

and private actors, coordinate among them, and act in real time to counter foreign

information campaigns.

o The JIATF should draw on existing authorities to create an operations center with

modern, AI-enabled digital tools and expert staff to expose, attribute, and respond

effectively.

o The Presidential action should also direct these officials, as part of the JIATF, to

create a mechanism to share and exchange critical information with key companies

in the private sector that run internet and social media platforms where malign

information proliferates.

274

BLUEPRINT FOR ACTION: CHAPTER 1

Action for the Secretaries of State, Defense, Justice, Homeland Security, and the Director

of National Intelligence:

•

Establish the JIATF and Operations Center.

o These agency heads should direct immediate development of a plan to create

the JIATF and operations center with a focus on identifying those offices, bureaus,

and divisions within their agencies and the broader IC and law enforcement

establishment that are essential to the mission of countering foreign-sourced

malign information.

■ As part of this effort, the JIATF should leverage the authority provided by

Congress in the FY2020 NDAA to stand-up a Foreign Malign Influence

Response Center within ODNI.³

■ Components that will be critical to the JIATF include, among others, the

Central Intelligence Agency’s Open Source Enterprise and the National

Counterintelligence and Security Center.⁴ Leadership will need to ensure

involvement of relevant components from the FBI, the National Security

Agency, across the Department of Defense, and the Global Engagement

Center (GEC) at the Department of State.

•

The JIATF would lead and integrate existing and new national strategic efforts against

foreign malign information operations by providing analysis, sharing information with

government and commercial partners, and driving whole-of-government action,

subject to Presidential direction, to advance U.S. information objectives.

•

The Commission proposes that the operations center component of the JIATF be

modeled on the National Counterterrorism Center (NCTC), as a proven model for

providing real-time situational awareness of and response to evolving national security

threats.

•

To exchange information and coordinate with internet and social media platforms on

malign information threats, the Commission proposes creation of an associated industry

consortium that includes an information sharing and analysis center (ISAC). The

consortium, supplemented by the ISAC, would allow the JIATF to exchange information

with industry, monitor malign information across ICT platforms, and improve U.S.

government response to malign information threats. In developing the ICT consortium

and ISAC, JIATF should look to the Global Internet Forum to Counter Terrorism as a

model.⁵

Action for the Director of National Intelligence:

• Appoint a Malign Information Threat Executive (MITE) to lead the JIATF.

o In July 2019, ODNI created the Election Threat Executive position responsible

for coordinating across the IC on issues related to election security.⁶ The threat

of foreign malign information operations demands that this position be elevated,

renamed, and expanded beyond the subject of elections.

o The MITE role should also serve as a liaison function between the White House/

National Security Council and the JIATF to ensure alignment and responsiveness to

the national security strategy.

275

EMERGING THREATS IN THE AI ERA

Action for Congress:

• Appropriate $30 million per year to support the operations of the JIATF.

Operational Framework

Efforts by the U.S. Government and private sector to counter terrorist propaganda offer a

potential roadmap for how the United States can go on the offensive to counter and compete

against malign information. The creation of the Global Coalition to Defeat the Islamic State

of Iraq and Syria (ISIS) has shown how a burden-sharing model can be deployed to

successfully counter and defeat a shared threat.⁷ The United States and its allies will only

succeed if they can develop and deploy personnel as well as an advanced set of tools to

assist in their effort to counter and compete against malign information operations. Efforts

need to be made to encourage innovation as well as harness commercially available

technologies to go on the offensive.

Recommendation: The Department of State should lead a global effort to counter

Recommendation

disinformation.

Action for the President:

• Designate the Under Secretary of Public Diplomacy and Public Affairs at the

Department of State to lead the international fight against malign information

operations.

Action for the Department of State:

• Build an International Task Force to Counter and Compete Against Disinformation.

Modeled afer the Global Coalition to Defeat ISIS, the Department of State should

build a similar task force to counter malign information. The International Task Force

to Counter and Compete Against Disinformation should be led by the Department

of State’s Under Secretary for Public Diplomacy and Public Affairs, with the GEC

coordinating its daily activities.⁸ The task force will be in charge of directing, leading,

synchronizing, integrating, and coordinating efforts by allies to recognize, understand,

expose, and counter foreign state and non-state propaganda and malign information

efforts. The GEC should leverage the work of the Technology Engagement Team (TET)

to share and test technologies to detect and disrupt the creation, manipulation, and

dissemination of malign information from state and non-state actors. See the Chapter

15 Blueprint for Action for more detail on creating a task force as part of the Emerging

Technology Coalition proposed by the Commission.

Recommendation

Recommendation: The Defense Advanced Research Projects Agency (DARPA) should

coordinate multiple research programs to detect, attribute, and disrupt AI-enabled malign

information campaigns and to authenticate the provenance of digital media.

The government should sponsor research to develop technologies to detect, attribute,

and disrupt malign influence operations, including influence campaigns, psychological

operations on social media platforms, and manipulated and synthetic media. In parallel,

276

BLUEPRINT FOR ACTION: CHAPTER 1

the government should develop alternative technologies to authenticate the provenance

of digital media and head off the risk that other approaches will not be successful. These

efforts should be led by DARPA.

Action for Congress:

• Appropriate $60 million to $80 million in additional funding for DARPA to sponsor

multiple research projects to develop technologies to detect, attribute, and

disrupt malign influence operations that rely on AI-generated content, and to

develop alternative technologies to authenticate the provenance of digital media.⁹

DARPA has existing authority to fund such research with the scope outlined in this

recommendation, but will require dedicated appropriations to carry out the effort and a

security review of the best innovation vehicles to sponsor the research.

Action for DARPA:

• Sponsor further research as described above using innovation vehicles, such as

challenge competitions, or any other deemed necessary by DARPA to develop

and transition these technologies to accountable agencies and departments for

maximum employment.

Recommendation: Create a task force to study the use of AI and complementary

Recommendation

technologies, including the development and deployment of standards and technologies,

for certifying content authenticity and provenance.

In response to the challenges of misinformation, efforts are underway to develop standards

and pipelines aimed at certifying the authenticity and provenance of audiovisual content.¹⁰

These efforts make use of technologies, including encryption and fragile watermarking, to

secure and track the expected transformations of content via production and transmission

pipelines. These efforts offer the opportunity to mitigate malign information campaigns that

seek to corrupt or spoof highly trusted sources of information across our digital ecosystem.

This technology area is ripe for public-private partnership, as several private organizations

are already forming to fight disinformation.¹¹

Actions for the Office of Science and Technology Policy (OSTP):

• Establish a task force to study the use of AI and complementary technologies for

certifying content authenticity and provenance.

o OSTP should establish an interagency task force to assess the use of AI and

complementary technologies to certify content authenticity and provenance, to

include an evaluation of technical standards and production and transmission

pipelines.

o The task force should make recommendations on methods to improve content

certification, which may include public-private initiatives, legislation, and changes

to federal policy. In addition, the task force should assess options for federal

regulation of content certification by non-governmental organizations.

277

EMERGING THREATS IN THE AI ERA

Recommendation: Executive Branch departments and agencies should utilize Other

Recommendation

Transaction Authorities (OTAs), creative investing, and the Small Business Innovation

Research (SBIR) program to deploy capital to companies that offer technical solutions that

will assist the United States Government in identifying, countering, and defending against

malign information operations.

The U.S. Government has an array of mechanisms that are not currently leveraged to

deploy capital to companies that create strategic technology to unleash AI, machine

learning (ML), and associated technologies in this counter-information operations fight.¹²

Action for all U.S. departments and agencies:

• Explore the use of the SBIR program and OTAs to acquire technology solutions that

will assist the United States Government in identifying, countering, and defending

against malign information operations.

The United States must prepare for both the present and future threat of increasingly

automated and AI-enabled cyber conflict. The expanding threats of mutating malware and

AI-powered tools are combining with traditional cyber threats to automate, optimize, and

ultimately transform the precision, speed, stealth, scale, and effectiveness of cyber-attack

and espionage campaigns.¹³ To defend the U.S. from current and future cyber threats,

we must move to develop AI-enabled cyber defenses and to mitigate proliferating cyber

vulnerabilities.

Chapter 1: Emerging

Threats in the AI Era

Blueprint for Action: Number Two

Preparing for AI-Enabled Cyber Conflict.

Section 1: Developing AI-enabled defenses against cyber attacks.

Recommendation: Develop and deploy machine-speed threat detection and mitigation.

Recommendation

Detecting and reacting to unknown threats on a network is difficult, but not impossible,

for self-learning AI systems that have been trained to differentiate between normal and

anomalous network behavior.¹⁴ To address deficiencies highlighted by the SolarWinds

attack, autonomous defenses are needed to defend the U.S. Government’s systems.

278

BLUEPRINT FOR ACTION: CHAPTER 1

Actions for the Department of Homeland Security and Department of Defense:

•

Expand machine speed threat information sharing, behavior-based anomaly

detection, and cyber threat mitigation to all government networks containing

sensitive information and critical functions.

o DHS must improve the National Cybersecurity Protection System (NCPS), while

DoD must also accelerate its efforts to harness AI-enabled cyber defenses and

sensors. At a minimum, the objective of these new defenses should be to flag or

potentially block never-before-seen connections and communications missed

by currently deployed intrusion detection and prevention technologies such as

EINSTEIN.¹⁵ To fully take advantage of new capabilities, these defenses should also

aim to accelerate recovery from cyber attack by automatically generating courses

of action for federal agencies to assure secure continuity of operations. These

defenses should assist recognition of insider threats as well as externally launched

attacks, and use machine speed information sharing to prepare other public and

private networks to defend themselves against detected threats.

o DoD and DHS must also assess and mitigate security risks posed by introducing

and enhancing threat detection systems. These systems will require precautions

against their elevated system access being used to deliver malware or abused by

other cyber threats. AI-enabled system components designed to mitigate new and

unknown threats likewise will need defenses against adversarial techniques.

o To minimize cost overruns in altering a multibillion-dollar project, DHS should

reprogram $10 million to investigate the best means to accelerate and set up AI-

enabled threat detection systems. This study would be tasked to look for synergies

with existing intrusion detection sofware and infrastructure, seek to address any

remaining key deficiencies found by GAO in the National Cybersecurity Protection

System, and to develop a final budget proposal for Congress.¹⁶ This study likewise

should aim to address how previous intrusion detection systems failed to detect the

SolarWinds cyber attack.

Recommendation

Recommendation: Execute large, instrumented, and realistic tests to gather data and train

AI-enabled cyber defenses.

AI-enabled cyber defenses require training to recognize potential threats, and sensors

to detect them. By experimenting with larger networks in realistic conditions, the United

States can train more robust AI-enabled cyber defense capabilities.

Action for Congress:

• Fund the Defense Advanced Research Projects Agency (DARPA) to sponsor

additional secure, instrumented, and realistic research on AI-enabled cyber

defenses.

o DARPA funding should be increased by $20 million, to be divided between a

security review, and other programmatic costs for the additional research. DARPA

should be lef free to determine the structure of further research, with an innovation

vehicle such as a challenge competition or any other that DARPA deems necessary.

• Expand the National Institute of Standards and Technology AI testbed program.

279

EMERGING THREATS IN THE AI ERA

o For FY2021, NIST requested a $25 million increase, for measurement tools and

testbeds to accelerate the development and adoption of interoperable, secure, and

reliable AI technologies.¹⁷ Since then, NIST has been authorized for $64 million in

additional AI R&D responsibilities including AI testbeds. To ensure NIST can meet

its new responsibilities in addition to its prior ones, Congress should meet NISTs

authorized funding increase for AI R&D.

Actions for DARPA:

•

Structure and standardize an innovation vehicle, such as a challenge competition,

or any other DARPA deems necessary, to increase insight about options for new AI-

enabled cyber defenses.

o DARPA should aim to encourage the prototyping of new means of AI-enabled cyber

defense and test the efficacy of these defenses against intelligent opponents and

AI-enabled cyber threats. DARPA should structure new research to broaden insight

on the importance of real-life factors such as cyber-attack externalities, differences

in risk tolerance between threat actors, and differences in network infrastructure

between defenders.¹⁸

•

Bring broader fields of expertise to bear for cyber defense research.

o Cyber expertise is not the only expertise relevant to cybersecurity and the efficacy

of cyber operations.¹⁹ The new research should involve experts from other fields

such as economics, game theory, and behavioral psychology to improve scoring

metrics, improve the human components of cyber strategy, and propagate insight

further within government. With these improved metrics and insights, future

investments can be more directly aligned with mission assurance.

•

Conduct a security review to determine the rules and bounds of new cyber

research initiatives.

o DARPA must conduct a thorough security review about the second-order effects

of sponsoring research with public-facing results and without strong information

security measures, to mitigate against potential adversaries acquiring information

that can be weaponized against us. International competition in this area is getting

so intense that the organization must consider using a vetted closed-challenge

competition or initiative as opposed to an open-challenge competition format.

Actions for NIST:

• Expand the NIST AI testbed program to generate data for AI-enabled cyber

defenses in differing IT infrastructure environments.

o Larger-scale testing is necessary to generate the data required for AI-enabled cyber

defenses. By scaling testbeds within NIST, there will be the opportunity to generate

this data, and to evaluate the performance of varying network architectures at

strengthening network security.

o Training data ofen reflects a broad sampling of common scenarios and does not

itself necessarily convey the costs of different types of compromises without further

labeling.²⁰ NIST should create optimized data sets for training cyber defenses to

minimize expected costs of network disruption, compromise, and damage rather

than merely trying to identify cyber threats and vulnerabilities with high accuracy.

To develop these data sets, NIST will need to hire or contract multidisciplinary talent

to develop better metrics.

280

BLUEPRINT FOR ACTION: CHAPTER 1

Recommendation

Recommendation: Ensure the robustness of AI-cyber defenses.

To make AI-based cyber defenses stronger, their supporting supply chains and data must

be defended, while the algorithms themselves must be protected from malware, trained

against adversarial techniques, and red teamed to the point of failure. This approach can

be found in the Chapter 7 Blueprint for Action.

Section 2: Ensuring resilience against AI-enabled cyber attacks.

Many of the defenses required to protect against AI-enabled cyber threats are also

required to defend against less advanced cyber threats. To provide this protection, the

Commission endorses specific Cyberspace Solarium Commission recommendations,

which are instrumental in enhancing U.S. defenses against AI-enabled cyber threats.²¹

Recommendation: Improve incentives for information and cyber security.

Recommendation

AI cannot defend inherently indefensible digital infrastructure against escalating offensive

AI-enabled cyber capabilities. Even if vulnerabilities are known and easily patchable,

that is no guarantee that they will be closed without a further impetus to action. Similarly,

while new instrumented digital infrastructure is required to accelerate AI-enabled cyber

defenses, those that build it must be careful to ensure new vulnerabilities don’t outweigh

the benefits of these defenses. In both cases, incentives must be realigned in the public

and private sector to assure gaps are closed and new infrastructure is secure.

Action for Congress:

• Establish liability for final goods assemblers for damage stemming from incidents

that exploit known and unpatched vulnerabilities, incentivize reporting, and

amend the Sarbanes-Oxley Act to include cybersecurity reporting requirements.²²

o The Cyberspace Solarium Commission made recommendations to incentivize

timely vulnerability patching. In addition to these recommendations, companies

should be incentivized to improve their cybersecurity, and participate in new

vulnerability disclosure programs via selectively reducing legal liability and product

recalls for companies that can mitigate and patch controlled vulnerabilities within a

limited, but rule-defined, time period. The overall structure of liability reform should

aim to minimize perverse incentives to avoid liability by concealing failure. Grid,

critical infrastructure, and medical device companies should be the primary targets

for improvement.

o To harmonize and clarify cybersecurity oversight and reporting requirements for

publicly traded companies, Congress should amend the Sarbanes-Oxley Act to

explicitly account for cybersecurity.²³

281

EMERGING THREATS IN THE AI ERA

Action for the Executive Branch

• Incentivize information technology security through Federal Acquisition

Regulations and Federal Information Security Management Act authorities.²⁴

o Zero-trust networking and robust code should become key priorities for government

contracts related to information technology, and especially for contracts related to

AI. Contractors should not be paid more for additional lines of code when adding

them generates new vulnerabilities without additional functionality. Code should be

subjected to AI-enabled vulnerability review.

• Task CISA to develop an IT infrastructure “Cash for Clunkers” incentive plan, to

submit to Congress for FY2022.

o This program would support the replacement of vulnerable outdated equipment

with modern alternatives through targeted federal subsidies. CISA should

coordinate the effort by setting the program’s strategy, prioritizing devices and

critical digital infrastructure for replacement, and determining subsidy levels for

the systems to be replaced. CISA must develop the plan so as to minimize perverse

incentive to acquire vulnerable infrastructure before the plan is funded, and once

the plan is developed, Congress must implement it as quickly as possible to reduce

perverse incentives for companies to hold out on replacing vulnerable devices and

infrastructure in the meantime.

Section 3: Disrupting adversary AI-enabled cyber-attacks and capabilities.

Recommendation: Develop additional, impactful non-kinetic options to respond to

Recommendation

adversarial cyber and information operations.

Modern information operations have enormous overlap with cyber operations. As AI-

enabled cyber capabilities spread in the presence of wide-open societal vulnerabilities,

the United States needs to have additional tools to counter proliferating threat actors, and

to establish deterrence in the cyber and information domains.

Action for Congress:

• Expedite the establishment of the Bureau of Cyberspace Security and Emerging

Technologies (CSET) within the U.S. Department of State.

o The CSET Bureau will be essential for strengthening norms in cyberspace,

engaging other countries on information technology standards, assisting

allied cyber defense, and improving international cyber law enforcement.

Recommendations to expedite the Bureau’s buildout and ensure that it has a clear

mandate to coordinate strategy on the full range of emerging technology issues, in

addition to critical cybersecurity needs, can be found in the Chapter 15 Blueprint for

Action.

• Strengthen the U.S. Government’s ability to take down botnets by enacting Section

4 of the International Cybercrime Prevention Act.²⁵

282

BLUEPRINT FOR ACTION: CHAPTER 1

o Botnets are already a present threat, and may become more powerful with advances

in AI, not just directly spreading malware, but harvesting both computational power

and data to put toward further offensive training in ways that were not previously

possible. To enable the U.S. Government to better work with private industry and

international partners, Congress, in consultation with the Department of Justice,

should enact Section 4 of the International Cybercrime Prevention Act.²⁶ This

legislation would provide broader authority to disrupt all types of illegal botnets, not

just those used in fraud.²⁷

Actions for Cyber Command, the Department of Homeland Security, the Federal Bureau

of Investigation, and the National Security Agency:

• Expand current cyber threat inoculation initiatives.

o Machine speed information sharing is a key piece of enabling AI-cyber defenses.

To contribute to the readiness of U.S. defense and critical infrastructure, efforts

should be made to accelerate sharing of the most recent malicious code captured

in the wild through appropriate interagency channels, including through a Joint

Collaborative Environment. U.S. Cyber Command should ensure and accelerate

coordination with DHS, the FBI, NSA, and stakeholders in the private sector in the

release of threat information, particularly with owners and operators of systemically

important critical infrastructure.²⁸

Section 4: Coordinating and Strategizing a Response.

Recommendation

Recommendation: Reform the U.S. Government’s strategy, structure, organization, and

authorities for handling AI-enabled cyber threats.

The U.S. must organize and align authorities to fully implement the cyber security mission

and fully capitalize on machine speed information sharing defenses. Technology alone

isn’t enough: Cyber threat intelligence, joint planning, and response must be integrated

into the same organization to keep pace with AI cyber threats.

Actions for the Executive Branch:

• Issue an updated National Cyber Strategy with the following components.

o First, the strategy should build on the layered deterrence framework put forward by

the Cyberspace Solarium Commission with a focus on making the framework more

robust against the ways AI will transform cyber conflict.²⁹

■ To support the strategy, the Department of Defense, in partnership with the

Department of State and the IC, should also develop a multitiered signaling

strategy and promulgate a declaratory policy that addresses the use of AI in

cyber operations.³⁰

o Second, to inform the strategy, the Department of Homeland Security should run a

study to develop regulatory recommendations for the most cost-effective means of

defending digital devices and infrastructure. This study should investigate, but not

be limited to:

283

EMERGING THREATS IN THE AI ERA

■ Standards requiring critical private and public sector networks to keep their

data encrypted at rest and in transit

■ Multifactor authentication requirements for critical private and public sector

networks

■ Air gapping requirements for select sensitive, but still unclassified, networks

■ Analog defenses for cyber physical infrastructure to prevent the most lethal

failures regardless of how much network access cyber attackers gain, or how

advanced their methods of attack become

■ Federated machine learning techniques that lower espionage and privacy risk

via enabling data to be partitioned or remain decentralized

■ Specialized, narrow purpose computation hardware that can’t be repurposed

by malware for attacks

■ Ways to harness AI to lock down and constrain hardware toward its intended

purpose on vulnerable networks that can’t yet be patched or replaced

■ Ways to use cloud computing and virtual machines to reduce vulnerability of

AI and cyber systems to advanced persistent threats

•

Accelerate the establishment of a Joint Cyber Planning and Operations Center,

modeled afer the National Counterterrorism Center.³¹

o This planning office under the Cybersecurity and Infrastructure Security Agency is

necessary to coordinate cybersecurity planning and readiness across the federal

government and between public and private sectors. To properly stand-up such a

collaborative environment, the Executive branch must submit to Congress a list of

authorities and data sharing issues that will require additional authorities or funding.

•

Develop and implement an information and communications technology industrial

base strategy.³²

o This strategy must increase support to supply chain risk management efforts, and

provide better defense to the hardware supply chains, data, and algorithms that

compose the “AI stack.”

Action for Congress:

• Establish a Bureau of Cyber Statistics to inform both cyber defense policy and AI-

enabled cyber defenses.³³

o Large accurate data sets with relevant data are especially useful for training AI-

enabled cyber defenses that minimize the costs of cyber attacks and false alarms,

rather than just the number of attacks and false alarms. To that end, Congress

should establish a Bureau of Cyber Statistics, within the Department of Commerce,

or another department or agency, that would act as the government statistical

agency that collects, processes, analyzes, and disseminates essential statistical data

on cybersecurity, cyber incidents, and the cyber ecosystem to the American public,

Congress, other federal agencies, state and local governments, and the private

sector.³⁴

284

EMERGING THREATS IN THE AI ERA

• Authorize, establish, and fund a joint collaborative environment for sharing and

fusing threat information.³⁶

o Sharing and fusing threat information is an instrumental step in improving the speed

and capability of potential AI-enabled cyber defenses.³⁷ Congress must ensure that

Executive branch agencies have necessary authorities to bring their data together

in support of these efforts. Likewise, Congress must create incentives—including

liability protection—to attract the private sector to participate in threat information

sharing programs.

o To achieve these goals, the Commission endorses the Cyberspace Solarium

Commission recommendation for Congress to establish a ‘Joint Collaborative

Environment,’ a common, cloud-based environment in which the federal

government’s unclassified and classified cyber threat information, malware

forensics, and network data from monitoring programs are made commonly

available for query and analysis—to the greatest extent possible.³⁸

286

BLUEPRINT FOR ACTION: CHAPTER 1

Blueprint for Action: Chapter 1 - Endnotes

¹ For the purposes of this section, “malign information” includes both disinformation—false information

or intentionally misleading facts communicated with the intent to deceive—and misinformation—

false information not necessarily meant to deceive. See Daniel Fried & Alina Polyakova, Democratic

Defense Against Disinformation, Atlantic Council at n.1 (Feb. 2018); https://www.atlanticcouncil.org/

wp-content/uploads/2018/03/Democratic_Defense_Against_Disinformation_FINAL.pdf. For a broader

discussion, see Laura Rosenberger, Making Cyberspace Safe for Democracy, Foreign Affairs (May/

June 2020), http://www.foreignaffairs.com/articles/china/2020-04-13/making-cyberspace-safe-

democracy. For a study of how AI might be used to counter disinformation, see William Marcellino, et

al., Human-machine Detection of Online-based Malign Information, RAND Europe (2020), https://www.

rand.org/content/dam/rand/pubs/research_reports/RRA500/RRA519-1/RAND_RRA519-1.pdf.

² National Security Strategy of the United States, The White House at 34 (Dec. 18, 2017), https://

trumpwhitehouse.archives.gov/wp-content/uploads/2017/12/NSS-Final-12-18-2017-0905-2.pdf.

³ Pub. L. 116-92, The National Defense Authorization Act for FY2020, 133 Stat. 1198, 2129-30 (2019).

⁴ National Counterintelligence and Security Center, Office of the Director of National Intelligence (last

accessed Feb. 8, 2020), https://www.dni.gov/index.php/ncsc-home.

⁵ About, Global Internet Forum to Counter Terrorism (last accessed Oct. 2, 2020), https://www.gifct.

org/about/.

⁶ Press Release, Office of the Director of National Intelligence, Director of National Intelligence Daniel

R. Coats Establishes Intelligence Community Election Threats Executive,

(July 19, 2019), http://www.

dni.gov/index.php/newsroom/press-releases/item/2023-director-of-national-intelligence-daniel-r-

coats-establishes-intelligence-community-election-threats-executive.

⁷ Brett McGurk, America Should Build an International Coalition Now, The Atlantic (March 29, 2020),

https://www.theatlantic.com/ideas/archive/2020/03/america-should-build-international-coalition-

now/608983/.

⁸ Though this overall Blueprint for Action uses the term “malign information” to broaden beyond

disinformation to include misinformation, it will likely be easier to organize a task force around

countering disinformation.

⁹ Funding level should depend upon the number of programs DARPA has the capacity to execute in

this area.

¹⁰ See, e.g., Paul England, et al., AMP: Authentication of Media via Provenance, arXiv (June 20, 2020),

https://arxiv.org/abs/2001.07886.

¹¹ See Creating the Standard for Digital Content Attribution, Content Authenticity Initiative (last

accessed Feb. 19, 2020), https://contentauthenticity.org/; Overview, Project Origin (last accessed

Feb. 19, 2021), http://www.originproject.info/about.

¹² These could be SBIRs, OTAs, or other modern vehicles with minimal red tape. Recently published

reports on countering malign influence have issued wide-ranging recommendations including:

deploying special operations forces to areas “vulnerable to political warfare,” building “rapid-

reaction information cells to track and counter” malign influence operations, and promoting civil

society to “combine the values of accurate media with engagement skills and an understanding of

how propagandists prey on polarization, inflaming divides.” These recommendations are already

being put into action by Special Operations Command in Africa, using commercially available

services to combat and attribute malign information operations about COVID-19 and other issues

on the continent. The General Services Administration has awarded IST Research a Phase III SBIR

contract to “support operations in the information environment for the special operations and Federal

Government community.” Additionally, the U.S. Air Force and U.S. Special Operations Command

have contracted with Primer to “automatically identify and assess suspected disinformation” using ML

technology. See David Ronfeldt & John Arquilla, Whose Story Wins: Rise of the Noosphere, Noopolitik,

and Information-Age Statecraft, RAND at 72 (July 2020), https://www.rand.org/content/dam/rand/

pubs/perspectives/PEA200/PEA237-1/RAND_PEA237-1.pdf (citing or quoting other experts or

reports); Dave Nyczepir, SOCOM Looks to Combat Disinformation in Africa on New Governmentwide

Contract, FedScoop (July 27, 2020), https://www.fedscoop.com/socafrica-disinformation-ist-research/;

IST Research Awarded Five-year, $66 Million GSA Contract, IST Research (July 23, 2020), http://www.

287

EMERGING THREATS IN THE AI ERA

globenewswire.com/news-release/2020/07/23/2066650/0/en/IST-Research-Awarded-Five-year-66-

Million-GSA-Contract.html; SOCOM and US Air Force Enlist Primer to Combat Disinformation, Cision

PR Newswire (Oct. 1, 2020), https://www.prnewswire.com/news-releases/socom-and-us-air-force-

enlist-primer-to-combat-disinformation-301143716.html.

¹³ Nicholas Duran, et al., 2018 Webroot Threat Report, Webroot (2018), https://www-cdn.webroot.

com/9315/2354/6488/2018-Webroot-Threat-Report_US-ONLINE.pdf; Gary J. Saavedra, et al.,

A Review of Machine Learning Applications in Fuzzing, arXiv (Oct. 9, 2019), https://arxiv.org/

pdf/1906.11133.pdf; Isao Takaesu, Machine Learning Security: Deep Exploit, GitHub (Aug. 29, 2019),

https://github.com/13o-bbr-bbq/machine_learning_security/tree/master/DeepExploit; Catherine

Stupp, Fraudsters Used AI to Mimic CEO’s Voice in Unusual Cybercrime Case, Wall Street Journal

(Aug. 30, 2019), https://www.wsj.com/articles/fraudsters-use-ai-to-mimic-ceos-voice-in-unusual-

cybercrime-case-11567157402; Implications of Artificial Intelligence for Cybersecurity: Proceedings

of a Workshop, National Academies of Sciences, Engineering, and Medicine (2019), https://doi.

org/10.17226/25488; Ben Buchanan, et al., Automating Cyber Attacks: Hype and Reality, Center for

Security and Emerging Technology (Nov. 2020), https://cset.georgetown.edu/research/automating-

cyber-attacks/; Nektaria Kaloudi & Jingyue Li, The AI-Based Cyber Threat Landscape, ACM

Computing Surveys at 1-34 (Feb. 2020), https://dl.acm.org/doi/abs/10.1145/3372823; Dakota Cary &

Daniel Cebul, Destructive Cyber Operations and Machine Learning, Center for Security and Emerging

Technology at 5-23 (Nov. 2020), https://cset.georgetown.edu/research/destructive-cyber-operations-

and-machine-learning/.

¹⁴ Max Heinemeyer, Dissecting the SolarWinds Hack without the Use of Signatures, Darktrace (Jan.

7, 2021), http://www.darktrace.com/en/blog/dissecting-the-solar-winds-hack-without-the-use-of-

signatures/.

¹⁵ See EINSTEIN, U.S. Cybersecurity & Infrastructure Security Agency (last accessed Feb. 8, 2021),

https://www.cisa.gov/einstein.

¹⁶ Gregory C. Wilshusen, DHS Needs to Enhance Efforts to Improve and Promote the Security of

Federal and Private-Sector Networks. Government Accountability Office, (April 24, 2018), http://www.

gao.gov/assets/700/691439.pdf. See also Information Security: DHS Needs to Enhance Capabilities,

Improve Planning, and Support Greater Adoption of Its National Cybersecurity Protection System,

Government Accountability Office (Jan. 28, 2016), https://www.gao.gov/assets/680/674829.pdf.

¹⁷ President’s FY 2021 Budget Request to Congress for the National Institute of Standards and

Technology, National Institute of Standards and Technology (2020), http://www.nist.gov/system/files/

documents/2020/02/11/FY2021-NIST-Budget-Book.pdf.

¹⁸ Implications of Artificial Intelligence for Cybersecurity: Proceedings of a Workshop, National

Academies of Sciences, Engineering, and Medicine at 17-19 (2019), https://doi.org/10.17226/25488.

¹⁹ Implications of Artificial Intelligence for Cybersecurity: Proceedings of a Workshop, National

Academies of Sciences, Engineering, and Medicine at 12-19 (2019), https://doi.org/10.17226/25488.

²⁰ Implications of Artificial Intelligence for Cybersecurity: Proceedings of a Workshop, National

Academies of Sciences, Engineering, and Medicine at 15 (2019), https://doi.org/10.17226/25488.

²¹ Report, U.S. Cyberspace Solarium Commission (March 2020), https://www.solarium.gov/report.

[hereinafter CSC Report]

²² This recommendation modifies an existing Cyberspace Solarium Commission recommendation in

order to reduce the risk of creating perverse incentives to avoid enforcement. See recommendation

4.2 and 4.4.4, CSC Report at 76, 83.

²³ See recommendation 4.4.4, CSC Report at 83.

²⁴ See recommendation 4.4.3, CSC Report at 82.

²⁵ See recommendation 4.5.3, CSC Report at 87.

²⁶ S. 3288, International Cybercrime Prevention Act of 2018, 115th Congress (2018), https://www.

congress.gov/bill/115th-congress/senate-bill/3288/text.

288

BLUEPRINT FOR ACTION: CHAPTER 2

Chapter 2: Foundations

of Future Defense

Blueprint for Action

The Department of Defense (DoD) lags far behind the commercial sector in integrating new

and disruptive technologies such as Artificial Intelligence (AI) into its operations. Technical,

bureaucratic, and cultural challenges must be overcome to adopt AI to maintain the U.S.

military advantage. By 2025, the DoD must put in place the foundations for widespread AI

adoption, by: 1) Building the technical backbone; 2) Training and educating warfighters;

3) Accelerating adoption of existing digital technologies; 4) Democratizing development of

AI; and 4) Investing in next-generation capabilities. To the maximum extent possible, these

efforts should be coordinated with the Intelligence Community (IC) and other partners

across the national security community.¹

Recommendation

Recommendation: Drive Change through Top-Down Leadership.

Maintaining the defense advantage in an AI-enabled future will require top-down leadership

to overcome organizational barriers and create strategic change. Critically, civilian and

military leaders across the DoD and the IC must coordinate more closely, aligning priorities,

resources, and policies to speed technology adoption and research breakthroughs.

Actions for the Department of Defense and the Office of the Director of National Intelligence:

• Establish a Steering Committee on Emerging Technology, tri-chaired by the

Deputy Secretary of Defense, the Vice Chairman of the Joint Chiefs of Staff, and

the Principal Deputy Director of National Intelligence.²

o The Secretary of Defense and Director of National Intelligence should issue a

directive immediately establishing the senior oversight committee listed above.

o The Steering Committee on Emerging Technology provides a forum to drive

change, focus, and action on emerging technology that otherwise would not be

prioritized. It will enhance intelligence analysis related to emerging technology;

connect strategic vision to organizational change; focus concept and capability

development on emerging threats; guide defense investments that ensure

America’s strategic advantage against near-peer competitors; and provide the

authority to drive technology adoption and application by the Department.

• Assign the tri-chaired Steering Committee on Emerging Technology responsibility

for overseeing the development of a Technology Annex to the next National

Defense Strategy.³

291

FOUNDATIONS OF FUTURE DEFENSE

Action for Congress:

• In the National Defense Authorization Act (NDAA) for FY2022, establish a Steering

Committee on Emerging Technology and National Security Threats and designate

that it be tri-chaired by the Deputy Secretary of Defense, the Vice Chairman of the

Joint Chiefs of Staff, and the Principal Deputy Director of National Intelligence.

Recommendation: Build the Technical Backbone

Recommendation

Integration of AI into DoD operations requires urgent investment in a modern digital

ecosystem that will enable ubiquitous development and fielding at all levels—from the

headquarters to the tactical edge. It is essential to establish a technical foundation that:

1) Provides access to leading cloud technologies and services for scalable computing;

2) Enables the sharing of data, software, and capabilities through well-documented and

hardened application programming interfaces (API) with proper access controls; and 3)

Gives all DoD developers and scientists access to the tools and resources they need to

drive new AI capabilities. To this end, the figure below depicts the ecosystem managed

as a multilayer stack of services, accessed through common interfaces and providing

shared access to essential AI building blocks of data, algorithms, tools, trained AI models,

and compute. This should be realized through a federated approach, building on existing

resources and pathfinder efforts.⁴

The key elements that comprise the envisioned AI digital ecosystem are:

•

Contributors and Users. A diverse, distributed network that includes development

teams working at the tactical edge and at headquarters levels; private sector partners

contributing trained models and applications; academic researchers working on open

challenge problems; researchers working within a DoD lab; or international allies or

partners co-developing interoperable AI capabilities.

•

Common Interfaces. A service-oriented architecture where resources at each level

of the stack are accessed and maintained through common APIs based on industry-

standard protocols.

•

Authentication. Enhancing both the sharing and the safeguarding of resources through

a uniform policy and practice for managing authoritative, shared user attributes across

classification levels to control who will build, use, or share AI building blocks.⁵

•

Applications. Discoverable and accessible AI solutions ready for fielding through

provisioned platform environments.⁶

•

Platforms. Environments that support development, testing, fielding, and continuous

updating of applications to diverse sets of contributors and users.⁷ These platforms

include workflows and processes supporting the DevSecOps⁸ life cycle, MLOps⁹ for

machine learning pipelines, and digital engineering.¹⁰

•

Sofware. Federated sofware architecture¹¹ linking distributed repositories hosted across

the Department by mission components, their sofware factories, and service labs,

making sofware discoverable through a catalog.¹² Includes AI algorithms, data analysis

tools, and tools supporting test and evaluation, verification and validation (TEVV)¹³

as well as processes and tools to support continuous Authorization to Operate (ATO)

frameworks and reciprocity.¹⁴

292

BLUEPRINT FOR ACTION: CHAPTER 2

• Data. Federated and secured data architecture linking distributed repositories across

the department hosted by mission components, service labs, and enterprise programs,

making data discoverable through a catalog.¹⁵ With appropriate access controls, this will

facilitate finding, accessing, and moving desired data across the Department¹⁶ including

data sets, associated data models, and trained AI models along with supporting

documentation.¹⁷

• Hardware Infrastructure. Networking and communications backbone to transport

ecosystem resources, particularly data, and provide seamless access and interchange

between cloud computing and storage services.

AI Digital

Ecosystem.

OPERATOR

DEVELOPER

RESEARCHER

IN THE FIELD

AT AN OFFICE

ACCESS TO

ALL LEVELS

BY COMMON

INTERFACES

AUTHENTICATION

APPLICATIONS

Workflows

PLATFORM

DevSecOps

SOFTWARE REPOSITORIES

Algorithms

Tools

Datasets

Trained

AI Models

DATA REPOSITORIES

HARDWARE INFRASTRUCTURE

Network

Computing

Storage

Devices

293

FOUNDATIONS OF FUTURE DEFENSE

To accelerate the process of building on existing resources and pathfinder efforts, and to

increase interoperability in the short term, DoD should determine a governance structure

and develop necessary policies and guidance, draft a reference design, and make the

technical investments in the network and in platform environments. Implemented correctly,

the digital ecosystem will ensure force-multiplying common access and interoperability.

The Blueprint for Action framework outlined below marries top-down coordination and

direction with bottom-up mission implementation to realize an enterprise-wide ecosystem

in a manner that does not slow or stymy innovation, but rather incorporates new capabilities

at the speed of innovation and mission requirements.

Actions for the Department of Defense:

•

Establish Digital Ecosystem Leadership and Governance.

o The Secretary of Defense should direct the establishment of an enterprise-wide

digital ecosystem to support capability development to maintain the technological

superiority of the United States military.

■ To ensure senior leader oversight and sustained resourcing, the Secretary

should assign the Steering Committee on Emerging Technology the

responsibility to oversee the implementation and sustainment of the

ecosystem.

■ The Secretary should assign the DoD Chief Information Officer (CIO) as the

Executive Agent responsible for the ecosystem design, development, and

operation.

o The Steering Committee on Emerging Technology, coordinating with the DoD

CIO, DoD Comptroller, Director of Cost Assessment and Program Evaluation, and

appropriate acquisition and programming representatives from the military services,

should produce a funding plan¹⁸ for the ecosystem within 90 days of the Secretary’s

direction.

o The DoD CIO should form and chair an enduring digital ecosystem implementation

working group¹⁹ to establish and maintain an open architecture, an evolving

reference design, governance structure, and processes to include management

and authorization for ecosystem functions and growth. The Steering Committee

on Emerging Technology will ensure strategic direction and coordination,

and pathfinder organizations will provide bottom-up and mission-oriented

implementation.²⁰

■ The working group should report to the Steering Committee on Emerging

Technology, add members when appropriate, and include representatives

from:²¹

• The Office of the DoD Chief Data Officer (CDO)

• Component CIOs and CDOs

• The Joint Artificial Intelligence Center (JAIC)

• The Office of the Under Secretary of Defense for Research & Engineering

(OUSD (R&E))

294

BLUEPRINT FOR ACTION: CHAPTER 2

• The Office of the Under Secretary of Defense for Acquisition &

Sustainment (OUSD (A&S))

• The Office of the Under Secretary of Defense for Intelligence & Security

(OUSD (I&S))

• Service Acquisition Executives

• The Office of the Director of Operational Test and Evaluation (DOT&E)

• The Director of the Defense Advanced Research Projects Agency

(DARPA)

• Digital ecosystem pathfinders, including but not limited to, the Air Force’s

PlatformOne, Kessel Run, Space CAMP, the Navy’s Black Pearl, the Army’s

CReATE, ADVANA, and the Army Futures Command Sofware Factory²²

•

Develop and Mandate Participation in a Digital Ecosystem Governed by an Open

Architecture and Reference Design.

o Within 12 months of the Secretary’s direction to establish the ecosystem, the DoD

CIO should work with the implementation working group to develop and publish an

open, interoperable architecture²³ built on common interfaces based on industry-

standard protocols along with an evolving reference design.²⁴

■ The open architecture and reference design should be owned by the DoD CIO

and reviewed quarterly and updated through the working group.

■ An unclassified version of the open architecture and reference design should

be published publicly for commercial capability providers.

o The Secretary of Defense should issue a memorandum that requires all new joint

and service programs to participate in the digital ecosystem and adhere to the open

architecture.²⁵ This should include a requirement that all existing programs develop

a plan to participate and become interoperable with the digital ecosystem wherever

possible by 2025.

■ Through the Joint Requirements Oversight Council (JROC), the Vice Chairman

of the Joint Chiefs of Staff (VCJCS) and USD (R&E)²⁶ should ensure that all

joint and service programs designed to meet joint capability needs adhere to

the digital ecosystem open architecture.²⁷

o The DoD CDO, acting in coordination with the DoD Data Council, should ensure

that the Data Strategy Blueprints for Action Annex developed by each Component

under the DoD Data Strategy adhere to the digital ecosystem open architecture.

o The USD (A&S) should update the guidance governing the formatting requirements

for deliverable data in contracts to be well-documented,²⁸ “non-proprietary formats

designed for interoperability.”²⁹

o The Steering Committee on Emerging Technology should lead an effort with the

IC to assess additional ways to accelerate implementation and leverage the digital

ecosystem, including designating service providers to proliferate applications

across the enterprise and make them available for integration into complex mission

solutions.³⁰ Wherever possible, the digital ecosystem’s open architecture should

leverage and interoperate with proven solutions from the IC such as the Information

Technology Environment recommended in Chapter 5 of this report.

295

FOUNDATIONS OF FUTURE DEFENSE

•

Establish a Strategic Data Node.

o The DoD CDO should make it a priority to create a linked, large-scale, cloud-based

data repository (i.e., a node within the digital ecosystem) adherent to the data

service interfaces specified in the ecosystem’s open architecture. This would be a

critical step to enable distributed development efforts by providing AI development

teams secure access to authoritative data from diverse mission sets and functional

areas, and serve as a prototype for the digital ecosystem reference design.³¹

■ The CDO should create this strategic data node by integrating digital

ecosystem interoperability into the DoD ADVANA system³² and prioritize

construction of enterprise data sets as recommended below.

•

Expand the Network and Communications Backbone to the Digital Ecosystem.

o The Department should fully fund its network and communications modernization

effort as outlined in the DoD Digital Modernization Strategy,³³ require the DoD CIO

to factor this into their list of highest priorities, and hold the DoD CIO accountable

for expediting critical upgrades.

•

Create a Marketplace to Promote Democratization of AI Building Blocks.

o The DoD CIO, in accordance with the digital ecosystem governance and reference

design addressed above, should create an AI marketplace for strategic exchanges of

the essential AI building blocks.³⁴ The marketplace should include:

■ SofEx - GitLab-like³⁵ sofware repository system³⁶ hosting AI algorithms,

TEVV tools,³⁷ hardened AI sofware stacks, etc.

■ DataEx - a federated data repository system³⁸ of AI-ready data sets,

documentation, and associated data models.³⁹

■ ModelEx - a federated repository system of trained models⁴⁰ generated from

various types of AI approaches and techniques, including statistical machine

learning.⁴¹

■ CloudEx - a cloud-agnostic, networked marketplace for pre-negotiated

computing and storage services from a pool of vetted cloud providers.⁴²

o Trusted partners (inside and outside government) should be able to develop

solutions and products within secured environments of the ecosystem, offering

monetized access to users.⁴³

•

Develop Prototypical Platform Environments within the Digital Ecosystem.

o The DoD CIO should work closely with the digital ecosystem pathfinder community

to build a set of tailorable development environments for training AI systems

using: data-driven statistical machine learning; the latest simulation and modeling

capabilities to support reinforcement learning (e.g., digital twinning within an

accurate world model); and complementary TEVV services.⁴⁴

o The DoD CIO should work closely with the digital ecosystem pathfinder community

to implement a set of prototypical platform environments⁴⁵ that support

development, testing, fielding, and continuous update of AI-powered applications

for diverse categories of contributors and users.⁴⁶

296

BLUEPRINT FOR ACTION: CHAPTER 2

Action for Congress:

• Prioritize funding for the Department’s digital ecosystem and associated activities.

o The Armed Services Committees should use the FY2022 NDAA to direct the

Department to develop a resourcing plan for the digital ecosystem that establishes,

sustains, and incentivizes use of its various components as enterprise-wide,

enduring resources.

o The Committees should also authorize the obligation of funds to begin work on the

ecosystem.

Recommendation

Recommendation: Train and Educate Warfighters

Warfighters need the following capabilities to effectively build and use AI-enabled systems:

• Data-informed decision-making: Data-informed decision-making uses data to generate

insights and act on them. Data-driven organizations ofen make decisions more quickly,

at lower levels in the organization, and with a stronger empirical foundation than

organizations that rely primarily on intuitive or experience-based decision-making.⁴⁷

• Computational thinking: Service members need to better understand how to use

information processing agents to perform beneficial calculations that could not be done

quickly or efficiently by a human, rather than just representing human thinking in a

digital format.

• Maker culture: Service members of all ranks and occupations need regular contact with

AI-enabled machines, and should be able and encouraged to experiment with and

participate in the development of new tools.

• Human-machine teaming: Military leaders need to understand how to effectively

provide input to machines, interpret machine outputs, and critically, when to trust or not

trust machine outputs.⁴⁸

• Organizational transformation: Leaders need to understand when and how to integrate

AI-related tasks into their organization’s priorities, allocate resources needed to build

and maintain the AI stack, oversee the deployment and scaling of new systems, and how

to effectively interact with and support the careers of their technical experts.

Component 1: Integrate Digital Skill Sets and Computational Thinking into Military Junior

Leader Education.

Military junior leaders need to understand enough about AI to manage and operate AI-

enabled organizations responsibly and effectively. Commanding and leading AI-driven

systems and humans are very different fields. Leadership is even more complex in

organizations that combine human and AI elements. The below skill sets will equip junior

leaders with the fundamental skills needed.

Problem Definition and Curation. Military leaders need to understand problem curation,

or the process of discovering the causal mechanisms that lead to problems, associated

issues, stakeholders, and potential minimum viable products.⁴⁹ Poor problem definition

297

FOUNDATIONS OF FUTURE DEFENSE

and curation can lead to projects that attempt to solve the incorrect problem, wasting

significant amounts of time and money. This is particularly true for AI. Not all problems

can be solved using the type of probabilistic reasoning performed by many algorithms, or

with limited data sets. Also, many problems with potential AI solutions can be solved with

much easier, less-resource-intensive techniques. Military leaders that understand problem

curation will be better able to identify problems with potential AI solutions, and, just as

importantly, problems that AI will not help solve. This would not only help with the use of AI

but would also make junior leaders generally more productive.

A Conceptual Understanding of the AI Stack. The AI stack is a model that “provides a

streamlined approach to visualize, plan, and prioritize strategic investments in commercial

technologies and transformational research to leverage and continuously advance AI across

operational domains, and achieve asymmetric capability through human augmentation

and autonomous systems.”⁵⁰ A conceptual understanding of the AI stack would reinforce

the importance of building structural solutions to data collection, management, curation,

installation of sensors, and other underappreciated topics, and reduce attempts to add AI

at the end of a project. It will also help military leaders better understand what part of their

adversaries’ AI to target to degrade its effectiveness.

Data Collection and Management. Junior leaders need to understand how to collect

and manage data and to use systems that do the same in a manner that prepares it for

exploitation, and to operate in an environment where adversaries are doing the same. They

also need to understand the causes, effects, and ethical implications of data bias. Training

junior leaders to collect and manage data with the same degree of responsibility and

expertise that they use for medical care and equipment maintenance would accelerate the

government’s ability to create AI solutions, and to employ data-informed decision-making.

Understanding Probabilistic Reasoning and Data Visualization. Junior leaders need to

know enough about probabilistic reasoning and data visualization to understand the

outputs of their AI systems and their implications for a particular situation or environment.

This is critically linked to understanding when to trust and not trust a system’s outputs,

and other aspects of commanding and leading AI-driven systems. Notably, this does not

require leaders to perform computational statistics, just to understand their output, a much

less demanding task.

Data-informed Decision-making. To make data-informed decisions, leaders need to

understand system thinking and critical thinking. System thinking combines all of the

above to create an empirical but incomplete understanding of factors influencing a

decision, and how both their system affects their AI and how their decision will affect

their system. Critical thinking will help leaders understand the limits of AI, and the limits of

data-informed decision-making processes that are based on imperfect information. This

report references data-informed rather than data-driven decision-making because military

leaders should never be bound by the imperfect information in front of them. Their critical

298

BLUEPRINT FOR ACTION: CHAPTER 2

thinking, judgment, and intuitive understanding of both their system and their environment

will always have a critical role to play, even as it is informed by decision-making aids.

Action for Congress:

• Require the military services to integrate digital skills and computational thinking

into pre-commissioning and entry-level training.

o The Armed Services Committees should use the FY2022 NDAA to require the

military services to integrate understanding problem curation, the AI life cycle, data

collection and management, probabilistic reasoning and data visualization, and

data-informed decision-making into existing, pre-commissioning or entry-level

training for junior officers and training for non-commissioned officers within one

year of the passage of the legislation.

Action for the Military Services:

• Integrate digital skills and computational thinking into pre-commissioning and

entry-level training.

o The military services need to integrate understanding problem curation, the AI

life cycle, data collection and management, probabilistic reasoning and data

visualization, and data-informed decision-making into pre-commissioning or

entry-level training for junior officers and training for both junior and senior non-

commissioned officers. The military services can accomplish this by creating

new modules or courses, or by integrating this training into existing training and

education for commissioned and non-commissioned officers. Whenever possible,

this training should include the use of existing AI-enabled systems and tools.

Component

2: Integrate Emerging and Disruptive Technologies into Service-level

Professional Military Education.

While it is critical that military junior leaders better understand digital technology, military

leaders must also understand how technology will affect warfare, their roles in their

organizations, and how they should integrate new technology, both as they increase in

rank and responsibility and as technology changes.

Action for Congress:

• Require the military services to integrate emerging and disruptive technologies

into service-level Professional Military Education.

o The Armed Services Committees should use the FY2022 NDAA to direct the DoD

to require emerging and disruptive technologies courses for officers within one year.

The Armed Services Committees should also require the DoD to develop a training

plan that incrementally builds the necessary skills in its officer corps.

Action for the Military Services:

• Integrate emerging and disruptive technologies into service-level Professional

Military Education.

299

FOUNDATIONS OF FUTURE DEFENSE

o Course materials should address AI and other militarily significant emerging

technologies, as identified by the military services and the USD (R&E), in

coordination with the national laboratories, Federally Funded Research and

Development Centers (FFRDCs), and University Affiliated Research Centers

(UARCs).

o Course materials should include an introduction to the latest technology, the

benefits and challenges of adapting new technologies, how organizations

successfully and unsuccessfully adopt these technologies, and ethical issues

surrounding the uses of emerging technologies, including the impact of biases in

these technologies.

o As officers progress in rank, such courses should increasingly build the knowledge

base, vocabulary, and skills necessary to better understand new threats/challenges,

develop operational and organizational concepts, and incorporate technology into

operations and operational support.

o Military services should establish a mechanism that audits these courses annually

to ensure that emerging technologies have been properly identified, and that the

nomenclature, lexicon, definitions, and course content match changes in emerging

technologies.

Component 3: Create Emerging and Disruptive Technology Coded Billets in the Department

of Defense.

It is crucial that the DoD incentivize and increase the skill needed to introduce and field

emerging and disruptive technologies within the military officer corps. The joint qualification

process can serve as a model. The DoD already designates that certain critical billets must

be filled by Joint Qualified Officers⁵¹ and different levels of joint qualification.⁵² To do this,

the DoD should create emerging and disruptive technology designated billets for officers

that require an emerging and disruptive technology qualification prior to assignment and

a process for military leaders to become emerging and disruptive technology qualified.

Emerging and disruptive technology qualified officers would add value in a number of

areas for the services, including: 1) Assisting with acquisition of emerging technology; 2)

Helping integrate technology into field units; 3) Developing organizational and operational

concepts; and 4) Developing training and education plans.

Action for Congress:

• Require the Department of Defense to create emerging and disruptive technology

critical billets that must be filled by emerging technology certified leaders.

Actions for the Department of Defense:

• Create billets that require officers to become emerging and disruptive technology

certified before serving in the positions.

o The Office of the USD (R&E) should define emerging and disruptive technologies.

o Billets include, but are not limited to, positions that develop military doctrine and/

or operating concepts; positions within Force Structure, Resources, and Assessment

300

National Security Commission on Artificial Intelligence. Final Report - page 5