The US National Security Commission issues its "Final Report on AI in Defense and Intelligence" - here are the takeaways

I attended the Third Annual AI Summit at the Potomac Officers Club, a conference focused on AI for Defense and Intelligence.

The opening keynote by Yii Bajaktari, Executive Director, National Security Commission on Artificial Intelligence (NSCAI) covered the highlights of his recently released report with an uncomfortable message: America is not prepared to defend or compete in the AI era.

Meaningful decisions are needed now to accelerate AI innovation to benefit the US and defend against AI's malign uses. The oft-repeated trope from Henry Kissinger puts our current placement in this milieu: "When your scope for action is greatest, the knowledge on which you can base this action is always at a minimum. When your knowledge is greatest, your scope for action has often disappeared."  

The report proposes recommendations for President Joe Biden, Congress, and business and government leaders. The group points out that China is the first to challenge the US's technology dominance, threatening economic and military power since the mid-1900s.  

Expanding and democratizing AI research and development for a $40 billion investment is "a modest down payment for future breakthroughs," according to the 15-member commission, which suggests an investment from policymakers equivalent to the building of the interstate highway system in the 1950s. Longer-term, hundreds of billions of dollars of spending on AI are envisioned by the group.

The report makes far-ranging recommendations for business, tech, and national security. The backdrop of the global shortage of semiconductors proposes that the US stay "two generations ahead" of China in semiconductor manufacturing. It proposes a tax credit for semiconductor manufacturers. To address a global chip shortage and supply chain issues, President Biden pledged support for $32 billion.

"I hope that Congress deeply considers the report and its recommendations," AWS CEO Andy Jassy said in a meeting held to approve the report. "I think there's meaningful urgency to get moving on these needs, and it's important to realize that you can't just flip a switch and have these capabilities in place. It takes steady, committed hard work over a long period of time to bring these capabilities to fruition."

Here are the commissioners who helped to draft the report. Included in the report are:

• Oracle CEO Safra Catz, 
• Microsoft chief scientist Eric Horvitz
• Google Cloud AI chief Andrew Moore
• Andy Jassy, who takes over as CEO of Amazon later this year. 

Commissioners approved the 756-page report, calling for the United States to be AI-ready by 2025. "It bears repeating that to win in AI, we need more money, more talent, stronger leadership, and collectively we as a commission believe this is a national security priority, and that the steps outlined in the report represent not just our consensus but our conviction." Schmidt and the other commissioners said the work of selling these ideas to key decision-makers in power begins now.

Recommendations in the report include:

• Automation of many tasks by the intelligence community by 2030.
• The final report recommends hiring temporary or short-term tech talent and creating an accredited university to elevate government tech talent. Failure to do so is shortsighted and a national security risk.
• Open-source software for federal government employees in the Pentagon, pointing out that TensorFlow and PyTorch as "must-have tools in any AI developer's arsenal."
• In the next five years, private industry needs to invest in an organization with $1 billion in funding to launch efforts to address inequality.
• Identify service members with computational thinking.
• Allow for Director-level positions in each national security agency and the armed forces, charged with Responsible AI.
• U.S. State Department needs to increase its presence in technology hubs worldwide. 
• Increase by a factor of three AI research institutes.
• Set policy for the reporting of irresponsible AI deployments for agencies critical to national security.
• Double AI research and development.
• Immigration: commissioners add that doubling the number of employment-based green cards, creating visas for the makers of emerging and disruptive technology, for entrepreneurs, and giving green cards to every AI Ph.D. graduate from an accredited U.S. university. Referring to immigration as a "national security imperative," these measures could impede China's progress.

ACLU senior staff attorney Toomey said the report highlights some dangers of AI, but "It should have gone further and insisted that the government establish critical civil rights protections now before intelligence agencies and the military widely deploy these systems. Congress and the executive branch must prioritize these safeguards, and not wait until after dangerous systems have already become entrenched."

The report argues that the AI industry's consolidation threatens U.S. technological competitiveness in several fundamental ways, exacerbating brain drain trends and stifled competition.

Beyond Defense and Intelligence, the report adds extending recommendations to Congress for the FBI and border security and federal agencies such as the DHS. It raises federal agencies' issues as potentially affecting civil liberties and cites a lack of AI systems transparency. In an unexpected (for me) way, it calls for Congress to modify impact assessment and disclosure reporting requirements civil rights and civil liberty reports for new AI systems.

The report is focused mainly on China, as funding Increases by China to be an AI leader by 2025 part of its latest Five-Year Plan. The United States could lose its military-technical edge to China within the next decade. "We have every reason to think that the competition with China will increase," Schmidt said during the NSCAI meeting.

The report calls for the United States to establish an Emerging Technology Coalition with allies. The threat of rising models of authoritarian regimes like China's is real. AI presents opportunities to find areas for cooperation with China toward global challenges like climate change. Former White House economist and AI policy expert R. David Edelman said he recommended bilateral talks between the US and China.

A big surprise was that the commissioners argued against a treaty to ban AI-enabled autonomous weaponry. The reason stated that it is "not currently in the interest of U.S. or international security since Russia and China would not be bound by such a treaty," though an alternate was proposed to call for an international agreement. However, the report calls for an international agreement never to automate nuclear weapons and seek similar commitments from Russia and China.

The report recommends an executive order to protect intellectual property and create a Technology Competitiveness Council to codify intellectual property issues and national plans.

Oracle CEO Safra Catz described collaboration amongst DOD, U.S. government, and allies and said that government leadership is needed. "There's so many important steps that have to be taken now."

"It is our great hope that like-minded democratic nations work together to make sure that technologies around AI do not leak into adversarial hands that will give them an advantage over our systems and that we will unite together in the safe and responsible deployment of this kind of technology in military systems," commissioner and In-Q-Tel founder Gilman Louie said.

My take

It took me a while to realize that when they used the phrase "trustworthy AI," they did not mean it in the sense of being fair or equitable. Instead, "trustworthy," in the defense world, means AI that doesn't rain down on you, or turn around and blow you up.

What does this mean? AI systems will be used to gain and maintain power as weapons of first resort in future conflicts. It will not stay in the domain of the superpowers or the realm of science fiction. Adversaries are busy employing AI-enabled disinformation attacks to seed divisions in democracies and disrupt our comfortable reality. States, terrorists, and criminals will conduct AI-powered cyber attacks and pair AI software with commercially available drones to create smart weapons. 

The report adamantly puts forward that by 2025 the Department of Defense and the Intelligence Community need to be AI-ready. We need to win the AI competition that is intensifying strategic competition with China (and Russia, which has fewer resources than China but a significant bullpen of highly trained and intelligent engineers. China is a pure surveillance state, discourages individual innovation). China's plans, resources, and progress should concern all Americans. But China's use of AI domestically is a chilling pattern diametrically opposed to individual liberty. Its AI development as a tool of oppression and surveillance at home and increasingly abroad is a counterpoint to our goals.

We know advances in AI build on themselves and confer significant first-mover advantages. This report is over 700 pages of material, but between government reorganization and detailed information about AI, which is already familiar, a view of the Table of Contents (I added page numbers) provides a snapshot of the entire report of the breadth and depth of the report:

Part 1: Defending America in the AI era

Chapter 1: Emerging Threats in the AI Era - 43
Chapter 2: Foundations of Future Defense - 59
Chapter 3: AI and Warfare - 75
Chapter 4: Autonomous Weapon Systems and Risks Associated with Intelligence 89
Chapter 5: AI and the Future of National Intelligence - 107
Chapter 6: Technical Talent in Government - 119
Chapter 7: Establishing Justified Confidence in AI Systems - 131
Chapter 8: Upholding Democratic Values: Privacy, Civil Liberties and Civil Rights in Uses of AI for National Security - 141

Part 2: Winning the Technology Competition

Chapter 9: A Strategy for Competition and Cooperation - 157
Chapter 10: The Talent Competition - 171
Chapter 11: Accelerating AI Innovation - 183
Chapter 12: Intellectual Property - 199
Chapter 13: Microelectronics - 211
Chapter 14: Technology Protection - 223
Chapter 15: A Favorable International Technology Order - 241
Chapter 16: Associated Technologies - 253
Blueprints for action - 271 - 599

That's the backdrop for the 700 pages of the report, which is primarily concerned with "an integrated national strategy to reorganize the government. Reorient the nation and rally our closest allies and partners to defend and compete in the coming era of AI-accelerated competition and conflict."