ChinAI #280: Sour or Sweet Grapes? The U.S.'s Unstrategic Approach to the "Chip War"

Greetings from a world where…

grapes and gripes intermix

…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).

Sour or Sweet Grapes? The U.S.'s Unstrategic Approach to Chip Export Controls on China

Once upon a time, a Fox craved some grapes that hung from a tree’s high branches. First, the Fox jumped to try and reach the grapes. Then, he backed up to gain space and made a running leap. Over and over, he tried, falling short each time. Admitting defeat, he griped, “Here I am wearing myself out to get a bunch of sour grapes that are not worth gaping for.”

The typical moral that we derive from this fable is to not belittle something that is unattainable. In fact, this tale is the source of the expression “sour grapes.”

But perhaps this story also offers another lesson that directly bears on the U.S.’s approach to chip controls. There has been so much attention, time, talent, and other resources devoted toward figuring out how to implement these controls effectively. The Fox is building ladders, closing loopholes, and battling smugglers — all in an effort to reach those grapes. Yet, the basic question of whether those grapes are worth the effort has been largely neglected.

Allow me to present a different takeaway from the fable: before the Fox spends all its time trying various ways to reach those grapes, it should find out whether the grapes are sour or sweet in the first place. Again, to be clear, in this exercise, we are setting aside the question of whether the controls will actually work. Instead, we are simply exploring: if the controls are effectively implemented, will that actually be net beneficial for the national interest? So, let’s eat some test grapes.

First of all, there is significant confusion about the justifications behind the controls. The on-the-record rationale for the October 2022 controls is to slow down China’s military modernization efforts. Yet, whenever I bring up the fact that this betrays a fundamental misunderstanding of how AI will likely affect military power (think about how the most advanced military systems today are mostly dependent on legacy chips), supporters of the controls quickly pivot to two other reasons for the controls:

1. The U.S. needs to inhibit China’s overall technological capabilities (including the use of AI for commercial applications), since the U.S. must maintain a lead in foundational technologies.

2. The U.S. needs to curb China’s ability to develop artificial general intelligence (AGI), which reduces the potentially catastrophic risks of an AGI race.

Additional reason #1 represents a significant break in the U.S.’s approach to technological competition toward economic containment (the crudest form of this logic is that anything that hurts China’s economy is good for America). Additional reason #2 is never stated publicly, which limits open, public debate over the flimsy baseline assumptions.

Moreover, these controls carry significant disadvantages. For one, they significantly hamper international cooperation on AI safety (including catastrophic AGI risks). Imagine you are a moderate voice in Chinese leadership. Put simply, it is hard to take someone seriously when, with one hand, they preach about putting guardrails on AI development and, with the other hand, they seek to impair your economy. If one takes the need for international cooperation on AI safety issues seriously — as the current administration professes to do with its commitments at AI safety summits — it is important to recognize that these controls come with side effects.

The second disadvantage is that hurting China’s economy might not be in the U.S.’s best interest. When I have brought up this point — that a world with a weak China may be less safe than one with a stable China — to various audiences plugged into the administration’s thinking, they often react as if I have said something particularly provocative and radical. The truth is that a stable China has been a consistent objective of U.S. foreign policy for decades. As Mike Beckley, Associate Professor of Political Science at Tufts University, wrote in 2023:

Much of the debate on U.S. policy toward China concentrates on the perils of an ascendant and confident China. Yet the United States actually faces a more volatile threat: an ambitious and anxious China suffering a prolonged economic slowdown. China's economic growth rates are sliding and could fall further in the coming years as debt, demographic decline, environmental degradation, and other headwinds take their toll…Whereas a fast-growing China could afford to expand slowly and back down in crises—safe in the knowledge that its wealth, power, and status were rising and that the legitimacy of the CCP was secure—an economically sluggish China could be more desperate for economic outlets and primed to react violently to slights and setbacks.

We haven’t even mentioned other downsides such as the risk that the controls accelerate a) China’s development of an independent high-end chip capacity and b) the reorientation of chip supply chains in a way that insulates them from future controls. If you are operating in a world where AI does not make its mark until after at least a decade, these downsides become more salient and the benefits of the controls become more muted.

I certainly do not think that this is a slam-dunk case against the controls, but I do hope that this post sparks more debate over the policy’s benefits and drawbacks — beyond questions about their efficacy. Reasonable people can and should disagree, but it should be emphasized that it is impossible to have a grounded basis for democratic deliberation, if the rationale behind a policy constantly shifts. Last Thursday, the U.S. rolled out a new set of China chip controls (this time, with some quantum computing sprinkled in!).

Before we spend all our time tracking whether the Fox will be able to reach those high grapes, let’s take a step back and reevaluate our strategy. It’s possible that we have misread Aesop’s Fox as an envious curmudgeon. Maybe the real reason he finally gave up was because he realized that the grape juice was not worth the squeeze.

ChinAI Links (Four to Forward)

Should-attend: CSET book talk on Technology and the Rise of Great Powers

Thanks to CSET for hosting a webinar for my book this Wednesday, September 11, from 12-1PM eastern. I’ll give a brief talk, followed by a Q&A moderated by CSET Research Analyst Hanna Dohmen. Register to secure your spot!

Should-read: A Growing Yard: The Biden Administration’s China Export Controls Are Ensnaring CPUs

Alongside Jacob Feldgoise and Brian Love, Hanna recently published a great piece on how the export controls have (seemingly) unintentionally ensnared CPUs. This is based on analysis of over 900 AI chip products.

Should-apply: Frontier safety & governance post at DeepMind

DeepMind has an open Research Scientist position on its Frontier Safety and Governance team at Google DeepMind. They’re looking for scholars with experience in national security, international relations, international security and the global governance of AI. Application deadline is September 20th.

Must-read: Survey of Chinese Students on AI Risks

For China Brief, Nick Corvino and Boshen Li published a survey of over 1,000 students at Tsinghua and Peking University regarding their views on AI risks. Some of the key takeaways below:

Students are more optimistic about the benefits of AI than concerned about the harms. 80 percent of respondents agreed or strongly agreed with the statement that AI will do more good than harm for society, with only 7.5 percent actively believing the harms could outweigh the benefits. This, similar to other polling, indicates that the PRC is one of the most optimistic countries concerning the development of AI.

Students strongly believe the Chinese government should regulate AI. 85.31 percent of respondents believe AI should be regulated by the government, with only 6 percent actively believing it should not…The strong support for regulation in the PRC, even as optimism about AI remains high, suggests a distinct perspective on the role of government oversight in the PRC context.

Students ranked AI the lowest among all possible existential threats to humanity. When asked about the most likely causes of human extinction, misaligned artificial intelligence received the lowest score. Nuclear war, natural disaster, climate change, and pandemics all proved more concerning for students.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99