Mark Kennedy’s speech at the BGF Conference “AIWS – New Democracy”
Harvard University Loeb House, November 25, 2024
During my first year in Congress in 2001, I faced a vote on whether to maintain Normal Trade Relations with the People’s Republic of China (PRC). I shared the belief of many at that time that democracy and capitalism were on the march and that digital technologies would empower citizens and consumers to help both spread across the planet. We believed then that the internet would empower citizens to hold their governments accountable. It was inconceivable that innovation could flourish in an environment of restricted access to information like in the PRC. We hoped that greater engagement with China would integrate them into the rules-based international order. With this view of the future, I voted to continue China’s trade status.
Oops.
The passage of time has revealed three realities. First, technology has undermined democracy and empowered authoritarian regimes. Second, the PRC is intent on undermining the rules based global order. Third, closed societies like the PRC can advance technology.
In 2017, the PRC issued an artificial intelligence (AI) development plan seeking to establish China as the world leader in AI by 2030. In August of 2021, China surprised the US by testing a new hypersonic missile that circled the globe and then maneuvered to its target with its guidance and maneuvering likely using AI. This threat to the US homeland was enabled in significant part by American technology. Even so, it demonstrated how the PRC’s unwavering focus and immense government resources can achieve ambitious technology goals.
Inflection Point
The US is at an inflection point. Depending on our response, we will either retain our edge in AI and other technologies that underwrites our economic leadership and our military superiority or surrender it to an ascendant China at great risk to our future prosperity and national security.
A congressionally mandated bipartisan commission led by Eric Schmidt, former CEO of Google, and former Deputy Secretary of Defense Robert Work concluded in 2021:
“America is not prepared to defend or compete in the AI era. This is the tough reality we must face. And it is this reality that demands comprehensive, whole-of-nation action…to defend against AI threats, responsibly employ AI for national security, and win the broader technology competition for the sake of our prosperity, security, and welfare… China’s plans, resources, and progress should concern all Americans. It is an AI peer in many areas and an AI leader in some applications. We take seriously China’s ambition to surpass the United States as the world’s AI leader within a decade.”
America’s actions to preserve our leadership in AI should match the scale of impact on our way of life if we lose our tech edge. Here are ten steps that America urgently must take to preserve its leadership in AI.
- Build Trust
AI’s rate of adoption will be determined by people’s acceptance of the technology as much as by the pace the technology advances. In Edelman’s annual survey, global trust in AI companies fell to 53% from 61% five years ago. Trust in the US dropped more – by 15 percentage points and to a lower level, with only 35% trusting AI companies. AI must be implemented in a way that Americans trust and the world trusts more than that offered by competitors.
Earning trust begins with raising the level of society’s understanding of AI’s importance to US competitiveness and the risks of letting the PRC take the lead in AI. Education institutions must better prepare the workforce for the future. The safety net must efficiently address any job dislocations. Governments should leverage AI to deliver better public services, not to undermine civil rights. To the extent possible, companies and governments should put control into the hands of the users.
America should implement balanced regulations that encourage innovation while addressing risks like algorithmic bias and misinformation. Overly restrictive regulations risk allowing the PRC to gain the lead, giving autocracies an edge over democracies.
- Invest in America’s Capabilities
Sanctions to limit access to US technology has limits. Overcoming US sanctions on the company since 2017 and more recent advanced chip restrictions, Huawei’s new chip is allowing it to gain smartphone market share from Apple. The US must sprint to stay ahead.
Research. Concerned that the PRC is approaching the US in total research spending, the Chips and Science Act passed in 2022 authorized a significant increase in research support. Yet Congress is failing to appropriate the level of research funding authorized. Fully funding the authorized research spending is essential for the US to retain its innovative edge, including dramatically increasing funding to spur discovery in foundational and applied AI technologies.
Talent. The PRC graduates twice the number of Science, Technology, Engineering and Math (STEM) PhDs as the US. With the US ranking near the bottom in the percentage of graduates pursuing STEM fields, it has long relied on attracting the world’s best and brightest. Visas for Chinese students and academics have fallen by two-thirds since peaking in 2015. A recent surge in students from India is taking up some of the slack. It is important that policies and a welcoming environment continue to attract the global AI talent essential to US tech leadership while investing more in STEM education for American students.
Compute Advantage. More powerful semiconductors give America a critical AI advantage today. To keep ahead, America must gain a decisive lead in emerging technologies that promise faster and greener compute power like optical computing (using laser light waves), neuromorphic computing (modeled after systems in the human brain) and quantum computing (using subatomic particles).
Innovative Ecosystem. Creating an ecosystem where AI can thrive requires access to high-quality, diverse datasets that safeguard privacy and cybersecurity. Collaborations between tech companies, academia, and government accelerate AI breakthroughs. A vibrant venture capital market that fosters AI-focused startups is another American advantage it must maintain.
- Don’t Aid PRC’s Military
Not wanting to sell rope that may be used to hang it, the US implemented export controls for advanced semiconductors and chip-manufacturing equipment in October 2022. Bifurcating technology markets comes at a cost, but security must take priority.
Last August, the government implemented outbound investment restrictions on technologies with important security implications–semiconductors, AI, and quantum computing.
Access to restricted AI chips is still allowed through cloud-based services and advanced AI models. The US should apply restrictions in both categories, doing so in a way that preserves security while also ensuring global tech leadership.
The difficulty in implementing restrictions was evident in the mysterious appearance of TSMC tech in Huawei’s recent chips. Effective restrictions must evolve as technologies advance and workarounds are discovered. Since “More than nine-tenths of the time it takes an AI model to respond to a user query is spent shuttling data back and forth between logic and memory chips” restrictions on high-bandwidth memory (HBM) chips should be considered.
- Don’t Allow Things or People to be Manipulated
Things. The PRC’s Cybersecurity and National Intelligence laws compel companies to allow the government to secretly access data. The PRC’s recent tech crackdown reflects an even tighter control of private companies. These actions raise concerns about the ability of the PRC to spy or disrupt. This has motivated limitations on Huawei telecom equipment, port cranes, and software on vehicles. Further restrictions are being proposed for small drones and LiDAR sensors. The PRC is implementing its own restrictions.
The US needs to comprehensively survey the landscape and address all vulnerabilities where a competitor could access and control America’s things. The US needs to pay more attention to nations it might rely on to mobilize against PRC aggression. If PRC President Xi Jinping knows that Huawei’s heavy presence in the Philippines allows it access to detect US intent and complicate US action, it limits the deterrence impact of the expanded US military presence there.
People. The US is also rightly concerned about outside powers seeking to influence American public opinion whether it be through deep fakes or the algorithms of social media platforms. This is the logic behind the forced sale or ban of TikTok. The US needs to comprehensively address all platforms that present similar risks.
- Eliminate Dependencies in Tech Supply Chain
While continued trade ties between the US and the PRC helps deter aggression, Xi Jinping could be tempted to act if he perceived the US as so dependent for critical supplies on the PRC that America would be hamstrung if conflict arose.
This has led the US government to intensify its focus on supply chains and subsidize domestic semiconductor production. To achieve true resilience the US must do more to cultivate diversified supply sources for needed goods and services by working with other countries to develop authentically separate supply chains. This is particularly true for rare earth elements and ingredients essential for tech manufacturing.
- Market to Most Nations Possible
Restricting the PRC’s access of advanced chips will intensify its efforts to enhance its own semiconductor supply chain. Western chip equipment companies “are already seeing declines in revenue as they are forced to leave the Chinese market.” America risks losing it global market leadership that funds the research necessary to keep a tech lead if it does not proactively seek to include as many nations as possible in its tech ecosystem. The PRC is aggressively advancing its own AI exports with initiatives like the Digital Silk Road. The PRC will be happy to add as a customer any country the US excludes. Tech restrictions must seek to preserve security while permitting selling technology to the greatest number of nations possible.
The Gulf countries are an example of the choices America faces. They are offering huge subsidies and cheap energy to grow an AI industry. Saudi Arabia has committed to spend $100 billion on AI to keep up with its neighbor, the UAE. America has concerns sharing its most cutting-edge technologies given human rights concerns and their deep academic and business connections with the PRC. Yet America doesn’t want them to unambiguously join the PRC camp if denied access to US technology. The government’s balancing is reflected in Microsoft’s collaboration with the Saudi Data and AI Authority to advance innovation in the Kingdom’s Generative AI sector. These decisions must be carefully weighed and continuously monitored.
- Contest Access to Data
In the age of AI, those with the most data are advantaged. China’s data is generally not available to the US, while China can access many non-sensitive US data sets. America’s openness is its strength, but careful consideration must be given as to which datasets should be restricted.
Huawei has a 35% to 40% market share of equipment outside of North America. Is it finding opportunities to intercept data to be used for AI training without the owners’ knowledge? This is easiest to imagine where it offers its equipment as a managed service. There are similar concerns with the rapid expansion of Chinese cloud providers in Southeast Asia and Latin America. Chinese subsidies give it an edge. Data localization requirements and price sensitivity hamper US cloud providers. US must do more to promote trusted and secure data strategies.
The digital yuan, China’s central bank digital currency, is designed with the capacity to collect data on its users and transactions. This adds urgency to the contest over payment systems.
- Lead AI Standards
Global standards will either support AI systems that empower citizens or control them. The PRC has long sought to influence global standard setting bodies like the International Telecommunication Union. America must work closely with private industries and allies to ensure that international bodies set technical standards for AI that align with its values.
- Collaborate with Allies
Alignment on Tech Restrictions. Actions on limiting access by revisionist powers to technology will only work if allies with similar technologies don’t circumvent such restrictions. That is why the US has worked closely with Japan and the Netherlands in crafting restrictions on advanced chips and chip-making equipment. It would be impossible to effectively restrict high-bandwidth memory (HBM) chips without the collaboration of South Korean producers Samsung and SK Hynix. Aligning actions among nations is hard, particularly with significant commercial interests at stake. Staying aligned with allies and partners will be both more important and more difficult as the AI race progresses.
Innovating Together. Innovation productivity is fueled by iterative collaboration among those who specialize in a specific field. It is vital that the US and its allies tighten research collaboration to maintain an aggregate lead in critical technologies. US scientists currently collaborate twice as much with Chinese partners as they do with those from Europe. Yet Europe has much to offer. For example, the EU is a strong competitor to China in all quantum technologies, including post-quantum cryptography, where the US lags China. The US would also benefit from more proactively pooling resources, talent, and innovation for AI advancements with European and Asian allies.
- Stay ahead on Defense AI Applications
AI is transforming warfare as shown by Ukraine’s success with AI driven drones. Both the PRC and the US are investing heavily in AI-driven military applications. The applications of AI are pervasive across the military – empowering autonomous vehicles on land, sea and air; digesting data from satellites, drones and other sensors to detect objects, spot anomalies and identify threats; responding to cybersecurity threats in real time; optimizing battlefield strategies; improving logistics; delivering virtual training and simulation; pinpointing targeting; engaging in offensive and defensive electronic warfare; facilitating human-machine teaming; permitting more flexible orbital warfare.
President Joe Biden recently released a National Security Memorandum on Artificial Intelligence. It “unequivocally states” the government should use AI only in ways that “align with democratic values,” and “specifically requires agencies to monitor the risk AI systems can pose when it comes to privacy, discrimination and human rights”. Providing clear guardrails can help accelerate adoption of AI in the defense and intelligence communities.
Yet arduous procurement systems, political forces resisting efforts to drop legacy systems and delays due to repeated continuing resolutions slows the US military’s implementation of AI-powered systems it needs to stay ahead. The PRC benefits from building its military capabilities anew. It is critical that the US finds ways to integrate new technologies quicker and better than the PRC.
Close
US actions to secure AI leadership are too ad hoc. Success requires a plan that comprehensively addresses the above action steps and authorizes a streamlined interagency effort. Governance should address real risks addressing AI bias, privacy and safety. But the US must also stay full speed ahead. While keeping a close eye on the risks of AI, it is important to ensure that governance efforts don’t disadvantage the US.
Another vote I took during the spring of 2001 was to stop debating the design of the World War II Memorial, now on our nation’s mall, and proceed with its construction. Only by taking the necessary steps to win the AI race can US strength deter aggression and avoid adding any more memorials to our national mall.
