ChinAI #270: Intelligence Revolution or Scale Revolution?

Fudan Prof. Yin Li analyzes the transformative potential of genAI from an industrial history lens

Greetings from a world where…

the Indiana Fever have won consecutive games!

…***Can you help ChinAI reach 200 paid subscriptions as we near the 20k subscriber mark? Write it off as a business expense! 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).

Feature Translation: Intelligence Revolution or "Scale" Revolution?

Context: Seven of the ten choices received votes, but the clear winner of this Around the Horn edition was #7: Beijing Cultural Review’s April issue on AI. This week’s translation, one of the featured articles in that issue (link to original Chinese), asks: What does the history of three past industrial revolutions tell us about how AI will transform the economy and society? The author, Yin Li [李寅], is assistant professor at Fudan University’s school of international relations and public affairs. He published China's Drive for the Technology Frontier: Indigenous Innovation in the High-Tech Industry (Routledge, 2022).

Key Passages: Li argues that only technologies that exhibit “increasing returns to scale on capital investment” have the potential to evolve into large-scale general technologies, which ultimately trigger major changes in the production process.

• For Li, recent advances in generative AI may signal this paradigm shift from a specialized technology for localized automation to a large-scale general technology. He writes, “In the traditional information technology industry, programs based on standard rules can be regarded as a specialized technology for local automation, which is suitable for high-frequency, repetitive tasks. Once the number of ‘exceptions’ beyond the coverage of the rules increases, traditional information technology will face the problem of rapidly decreasing returns to scale…After entering the generative AI stage, the increasing returns to scale of general large models can continuously provide positive incentives for investment: on the one hand, by continuously investing in computing power and (text data) corpus, the accuracy and usage scenarios of large models are constantly improving, and the unit cost of using large models is constantly decreasing; on the other hand, the high fixed investment of large models requires large-scale output and application to share the cost, which requires an expanding market to rationalize large-scale investment.”

• Li argues that similar paradigm shifts in past industrial revolutions can provide “the best historical reference for us to judge the potential impact of AI,” even though AI models produce more information and services rather than industrial manufacturing outputs. He writes, “Although artificial intelligence is a new technological breakthrough, human society has experienced many paradigm shifts from partial automation to large-scale general production technology. We can find a large number of cases and clues in industrial history: from the power loom in the first industrial revolution, to the automatic assembly line in the second industrial revolution, to the electronic computer in the information technology revolution, all have had similar impacts on a large number of existing industrial technologies and social organizations.”

• When he reviews this shift that took place in the First Industrial Revolution, Li claims that the cotton textile industry did not enter mass production until the mid-1800s (with the introduction of the semi-automatic power loom in 1834 and improvements in steam engine technologies).

If this pattern holds, then the key question — according to Li’s essay — is “what kind of institutional arrangements can support a society to smoothly complete the leap forward toward general AI technology for large-scale information production?”

• One important factor is a nation’s ability to broadly share the prosperity brought by new technologies across all classes and social groups. He cites the Luddite movement as an indication that Britain failed to properly manage the relationship between technological change and societal adjustment. He also considers whether, in the future, backlash will come from programmers and professional workers (lawyers, consultants, finance folks, etc.): “the more automated information production and decision-making process brought about by the general AI model is very likely to harm the above two types of high-skilled workers, making them a social force against technological revolution.”

• From his interpretation of history’s lessons, one factor that is less salient is technology embargo policy. From the essay: “Since the Industrial Revolution, any developed country's attempt to hinder the international diffusion of large-scale general technology has been futile in the long run. For example, in the 19th century, Britain imposed a strict technology embargo on the cutting-edge technology of the time, the power loom, but such a blockade could only last for less than half a century. When American companies launched fully automatic power looms through independent innovation in 1895, they quickly occupied the world market, and a series of industrial technologies related to cotton textiles in Britain had no chance to develop.”

Ultimately, Li is optimistic about China’s chances to manage this paradigm shift toward general AI technology for large-scale information production. He argues, “The transformation towards general AI technology is a brand new challenge for China. As an emerging industrial society that has only completed full industrialization in the past two decades, to date, almost all new technologies introduced by China could only bring welfare improvements, so we do not have much experience in handling technological changes with redistributive effects. But precisely because China has no historical burden and the service industry has a relatively low share in China's economy, China may be more likely to create new institutional arrangements, use AI technology to serve a wider range of people, and lead the world into an era of large-scale production of information technology.”

FULL TRANSLATION: Intelligence Revolution or "Scale" Revolution? ——The Evolution of AI Technology and Industry from a Historical Perspective

ChinAI Links (Four to Forward)

Should-read: Thread on Pentagon’s anti-vax campaign to undermine China’s Sinovac in the Philippines

In this week’s edition of “maybe the most impactful thing you can do in government is to NOT do stupid shit,” Reuters uncovered this secret anti-vax campaign run by the U.S. military. I found C Nguyen’s thread about this report, based on their experience in Indonesia, to be very illuminating.

Should-read: Teacher’s Pet

Matthew Dagher-Margosian and Eliot Chen, for The Wire China, provide a lengthy, informative report on the state of AI education in China. Great details on iFlytek as well as how AI education companies are subject to the tides of Party decisions (e.g., China’s tightening restrictions on private, after-school tutoring services).

Should-read: A price war breaks out among China’s AI-model builders

From The Economist:

Price wars are ten a penny in China. The emergence of hundreds of lookalike companies seemingly overnight is pushing down retail prices of everything from electric vehicles to bike-sharing and bubble tea. The latest products to enter the ruinous fray are artificial-intelligence (ai) chatbots.

Should-apply: GovAI Job Opportunities

The Centre for the Governance of AI is recruiting for a few different roles: a three-month Winter Fellowship, a one-year Research Scholar position, and a Research Fellow position (two-year renewable contract).

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