Comment on “The Impact of Foundational AI on International Trade, Services and Supply Chains in Asia”

Abstract

Meltzer (2024) discusses the key challenges and risks of artificial intelligence (AI)/ChatGPT, as well as the role of international trade agreements and other economic fora in addressing them. Meltzer also argues that the less formalized grouping such as the US-EU Trade and Technology Council (TTC), the Indo-Pacific Economic Framework for Prosperity (IPEF), and the Quad can do a better job in this, than traditional trade negotiating agreements.

I have three comments, each from trade policy, Schumpeterian, and geopolitics perspective, respectively.

First, Meltzer points out that IPEF and US-EU TTC are not for trade negotiations but for aligning approaches to technological issues involving opportunities and challenges of emerging technologies. For instance, IPEF addresses cooperation on regulation and standards related to AI, whereas the TTC discusses trustworthy and innovative AI as its key priority. This is an important interpretation, given the existing criticism that IPEF does not offer any new (US) market access to member economies, and so its impact will be limited. However, it makes sense for the USA as it now perceives free trade agreements (FTA) having benefited not the USA but China. Whereas FTA have resulted in losses of manufacturing jobs in the USA, IPEF might not do so but rather contribute to securing future high-tech markets and global value chains (GVC).

Second, a Schumpeterian comment is about ChatGPT (generative pre-trained transformers), which is pointed out as GPT (general purpose technologies). Meltzer discusses the results of an interesting research that with the spread of ChatGPT, global gross domestic product (GDP) may increase by 14% by 2030, creating additional incomes of $US 16 trillion. Such positive impacts are possible because many job tasks can be finished faster with the same quality. In the literature, GPT are defined as technologies that are pervasive, exhibit strong complementary abilities, and thus can be used in a multitude of applications (Bresnahan & Trajtenberg, 1995). Some argue that the 4th Industrial Revolution (4IR) technologies might be GPT given their transformative potential involving digitization and automation. A similar term in Schumpeterian economics is “key factor inputs” which exhibit low and falling relative costs, with availability of supply over long periods and the potential for the use or incorporation in various products and processes throughout the economic system. Such inputs, once they appear, may correspond to a techno-economic paradigm shift (Dosi, 1982) and thus bring in a new long wave of boom in economies. Examples of key factor inputs were microelectronics since 1970, or electricity in the 20th century.

There are ways to verify whether or not some technologies are GPT. Lee and Lee (2021) use three criteria: (i) “generality” (e.g. technologies with wider applications and impacts and thus cited widely); (ii) originality or combinatory technologies which are combining technologies from diverse fields; and (iii) short technology cycle time (e.g. technologies citing recent patents only, which implies a more radical break from the past). If some technologies are GPT, they will show higher generality and originality and short cycle time. Lee and Lee (2021) find that AI technologies fail to pass this test as they are similar or inferior compared to typical 3IR technologies, such as personal computers, the Internet, and cell phones. It might be interesting to conduct some analyses to find out whether ChatGPT may qualify as a GPT or key factor inputs. This should be an important research topic, and if the answer is yes, then we may hope for another long-run boom owing to the power of ChatGPT.

My final comment from a geopolitics perspective asks what will happen if IPEF and TTC succeed in their goals of a coordinated development of AI technologies but remain closed against, and decoupled with, China. Then an emerging scenario is the existence of two separate AI-related GVC, namely Western versus Chinese ChatGPT. The economic benefits to each bloc would depend on how powerful and efficient each ChatGTP is. But the global benefits or such effects of a long-wave boom will be less under this scenario than under the scenario of a more integrated global GVC. Such a scenario of decoupled ChatGPT eco-system would be likely, given that Chinese authorities set out guidelines that its ChatGPT are supposed to reflect “socialist values,” and that any release of ChatGPT should be approved by the government. Then, one may infer that Chinese-driven ChatGPT cannot be suitable in Western markets, and, at the same time, Western ChatGPT cannot enter China market. In the meantime, there might be a possibility that some countries may try to exploit both markets, for instance, India with high capabilities in software and AI technologies. India is interesting as it is a member of both the Quad and the Shanghai Cooperation Organization (SCO). Also, there may be a fight for third markets, for example, in Africa or the Middle East.