Michael Frank
The United States’ postwar preeminence gave it incredible capability to impose its security interests on other countries who were eager to access the U.S. market. The definition of those interests in terms of a rules-based order – one that would protect the sovereignty of smaller countries and facilitate mutual economic growth – was a welcome deviation from historical norms, in which powerful states dictated terms to smaller ones and international economics was a zero-sum proposition.
This liberal international order laid the groundwork for globalization, lifting hundreds of millions out of poverty and largely eliminating war between great powers. It ended the Cold War in the United States’ favor, with the Soviet Union unable to match the West’s relentless progress.
But there was a catch: in exchange for reaping the economic benefits of alignment with the United States, countries would have to align with U.S. security interests. That meant signing up to the U.S. foreign policy program of anti-communism and global strategic competition with the Soviet Union.
U.S. economic power often enticed countries to forsake independent strategic calculation. The United States’ share of global GDP at the outset of the Vietnam War was 38 percent. Many countries got rich on trade with the United States. As the late economist Richard N. Cooper argued, the overwhelming empirical evidence from the time was “that communism did not work well economically – it did not deliver significantly higher standards of living to ordinary people – as became evident especially with the growing contrast between East and West Europe, between North and South Korea, and between the People’s Republic of China and Taiwan and Hong Kong.” China itself would famously split from the Soviet Union and align itself with the economically superior United States, with fantastic results.
Today, U.S. GDP is not the same weapon that policymakers wielded to such great effect in the Cold War. The U.S. market is still the world’s largest, and in 2022 reached its greatest share of the world economy since 2006. However, unlike in the Cold War, that position is not a sufficient condition to attract and retain would-be strategic partners, for two primary reasons.
First, China’s economy is roughly 70 percent of U.S. GDP, far greater than the high-water mark of the Soviet economy, estimated at 46 percent in 1970. Second, trade liberalization is a dead end, both despite globalization’s success and because of it.
Neither of the United States’ two political parties are interested in the obvious strategic play in trade – the Comprehensive and Progressive Trans-Pacific Partnership (CPTPP), which U.S. allies have embraced. The CPTPP is certainly in the same diplomatic tradition of the postwar liberal institutions. However, rejoining the pact, while sensible for U.S. strategy, would not be the kind of game changing move that the actions of the late 1940s represented. U.S. average weighted tariff rates are already historically low – just 1.5 percent in 2020, compared with nearly 60 percent before World War II. Trade is a well-worn playbook. As a result, there is little room to keep giving.
Instead, the United States has developed a new area of dominance that the rest of the world views with a mixture of awe, envy, and resentment: artificial intelligence. Recent improvements in the technology are inviting upward revisions of productivity forecasts, spurring growth in advanced economies that has been out of reach for decades.
From AI models and research to cloud computing and venture capital, U.S. companies, universities, and research labs – and their affiliates in allied countries – appear to have an enormous lead in both developing cutting-edge AI and commercializing it. The value of U.S. venture capital investments in AI start-ups exceeds that of the rest of the world combined. The United States dominates Stanford HAI’s latest AI Index Report, with more than double the next best country’s mark in the number of newly funded AI companies, total AI private investment, AI patents, AI repository citations, AI repository publications, and AI conference citations. U.S. firms are responsible for nearly two-fifths of the value in the global semiconductor supply chain. U.S. foreign policy should be rooted in this natural strength.
Shifting the diplomatic narrative from the archaic goods trade paradigm and toward AI diffusion is critical to advancing U.S. interests. Economic gains in the 20th century came from trade liberalization. In the 21th century, the primary driver of rising living standards for advanced economies will be AI-based productivity enhancements after a multi-decade slowdown.
The empirical evidence suggests U.S. AI is far more open and effective than the competition. Partner countries may not share U.S. security goals, but policymakers everywhere are pursuing economic growth, moving up the value chain, and managing aging demographics. The value of AI is obvious to these countries. Consider how Indonesia recently decided to award OpenAI chief Sam Altman the country’s first “golden visa” based on their view that he “may bring benefits to Indonesia.”
Policymakers should consider how to reinforce this competitive advantage on the world stage while also acknowledging that no one country can go it alone. So far, while export controls on AI hardware have constituted sound policy, their formulation and future highlight the limits of security-related commitments in a vacuum. The United States failed to achieve alignment on its October 2022 export controls ahead of time, despite months of extensive lobbying. The Netherlands, while mostly going along with controls on lithography tools needed to make the most advanced chips, is skeptical of broader U.S. controls.
Ahead of the 2024 U.S. presidential election, it is reasonable to question whether that coordination has staying power beyond the personal diplomacy of U.S. President Joe Biden. Other countries that occupy critical stages of the AI hardware value chain, such as Germany and South Korea, have yet to implement controls. If defense treaty allies are already skeptical of this trade, the U.S. has little hope of achieving the kind of transformational grand strategy that defined the Cold War through normative appeals alone.
Instead, the United States should expand its framing from a narrow focus on export controls to a broader view of participation in the AI economy – one that rests on research and tools dominated by U.S. entities. As in the Cold War, such a framing would be most effective when it positions the benefits of aligning with the United States at the forefront of the discussion, rather than focusing solely on security policies that have comparatively little upside in the minds of allied policymakers. In the near term, five goals should be paramount.
First, the United States must achieve strategic alignment with the European Union, not just certain constituent members. To that end, possible alignment with the EU AI Act should be a top consideration for the ongoing AI regulatory effort in Congress. Setting an explicit goal of aligning AI governance could help to preclude the strategic divergence that the West’s adversaries are actively seeking. The EU’s ambition to set global AI risk standards may be difficult to achieve alone, but the influence of a transatlantic consensus would be formidable.
Second, some important U.S. strategic partners have foundational technology policies that could limit the proliferation of AI products and services. For example, India, South Korea, and Vietnam have explicit data localization requirements that make investment difficult and are particularly obstructive to the proliferation of AI-based products and services, which rely on API calls from models based elsewhere. Engaging those partners to demonstrate the potential of U.S. companies to support AI development, while preserving protections intended to support user privacy, local competition and national security, should be a top priority.
Third, in addition to existing initiatives like the U.S.-EU Trade and Technology Council, the U.S.-India initiative on Critical and Emerging Technology (iCET), and the U.S.-U.K. Comprehensive Dialogue on Technology and Data, the U.S. AI policy community should seek to establish bilateral and multilateral dialogues with a broader group of countries. Track 1.5 and Track 2 dialogues can be a powerful tool for showcasing U.S. AI ingenuity, while providing the context to U.S. entities of how their capabilities fit into a network of like-minded countries. Technology policy dialogues that consider cross-border data flows, joint cybersecurity, research funding, and investment must be organized around AI rather than social media and e-commerce.
Fourth, the logical next step for dual-use, AI-related export controls should include AI APIs, products, and services. If an AI chip could conceivably be deployed in support of developing weapons of mass destruction, so too could a large language model accessed through an API.
Fifth and finally, separate from these initiatives, a top foreign policy priority should be mitigating existential risk and the harms of AI catastrophes. These conversations should take place regardless of strategic competition. Risk-focused researchers should be empowered to share information about threats they uncover from rogue, misaligned or self-replicating AI.
The consequences of getting this security and economic alignment correct are just as serious as they were in the 1940s. Those strategic moves led to an era of unprecedented international peace, rising incomes, and standards of living. However, it is important to remember that they were rooted in U.S. strengths. Today, trading U.S. AI capabilities for security alignment could initiate the evolution of the global liberal order underpinned by U.S. leadership, setting the stage for decades more of peace and prosperity with a virtuous cycle of integration and growth.
No comments:
Post a Comment