Nobel laureate Paul Romer sees diminishing returns for AI, as FDI still 'killer app' for emerging economies

The artificial intelligence (AI) industry is "going through a kind of a bubble right now" and could soon hit a ceiling when it comes to productivity gains, according to Nobel laureate and former chief economist at the World Bank Paul Romer.

"The growth of digital technology hasn't in the past ushered us into this new era of much more rapid productivity growth, and I doubt that it's going to do it now," Romer told the South China Morning Post on the sidelines of the UBS Asian Investment Conference in Hong Kong on Tuesday. "So I think the perception that somehow everything has changed is like the perception that cyber currencies are going to change the world."

An escalating tech war between the world's two largest AI powers, the US and China, has threatened to force emerging economies to pick a side, with ramifications for their future growth. AI development has proven significantly lopsided across different cultures and languages.

Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.

For Romer, though, waiting for AI to push worker productivity growth to new heights is a fool's errand. The surest way to growth, he said, is foreign direct investment (FDI).

"There's two different ways you can think about globalisation unfolding," he said. "One is Japan has companies that make cars, and then they put those cars on boats, and they ship them all over the world. Another is Japan has companies that are good at making cars, and then they go make cars in the United States and other places around the world. I think people underestimate, especially in the developing world, the gains that can come from that type of direct foreign investment."

The primary reason for that is how workers will gain new skills on the job, he said, and that has downstream implications for an economy. Without those work opportunities, people miss out on chances to build up that knowledge and experience.

The other challenge that Romer sees is that AI, as an algorithmic technology that is just "trying to fit a curve", is bad at predicting and understanding rare events that humans can easily comprehend. He used the example of autonomous driving, which works well in ideal conditions but can result in accidents in scenarios with more complicated traffic patterns or reduced visibility.

Bill MacCartney (right), chief technology officer at the venture capital firm SignalFire and computer science professor at Stanford University, discusses generative artificial intelligence at the UBS Asian Investment Conference in Hong Kong on May 29, 2024. Photo: Handout alt=Bill MacCartney (right), chief technology officer at the venture capital firm SignalFire and computer science professor at Stanford University, discusses generative artificial intelligence at the UBS Asian Investment Conference in Hong Kong on May 29, 2024. Photo: Handout>

Bill MacCartney, chief technology officer at the venture capital firm SignalFire and a computer science professor at Stanford University, suggested current AI technologies could already be on track to solve the issue of limited information.

"The limits to extrapolation to training data that [Romer] points to are primarily limits to particular approaches to modelling and not insurmountable limits that are never addressable even in principle," said MacCartney, who added that he had spoken with Romer about the topic at the conference. "Those kinds of challenges point to a need for continuing innovation in the methods of AI."

One sign that AI is advancing in this area is that large language models (LLMs) give better answers when asked to reason through a problem, according to MacCartney.

"The more we're able to introduce those intermediate steps of reasoning and understanding and interpretation, the better able our models will be at dealing with unforeseen situations," he said.

Since US-based start-up OpenAI's launch of ChatGPT in 2022, China has been scrambling for its own home-grown AI champion. Chinese firms have produced more than 200 LLMs, with the largest tech giants jostling for dominance in a domestic market walled off from foreign competition.

"There's been some exploration of other languages, but probably not of the vast amount of literature that's in Chinese," Romer said. "So there may still be some room [to advance AI] that could come from combining those two different bodies of data."

Business models around this type of AI have yet to be proven, however. Meanwhile, Beijing is grappling with slowing economic growth, which is now largely dependent on exports. The central government has been pursuing what it calls "dual circulation", which involves boosting domestic demand for locally produced goods while continuing to supply international markets as much as possible.

Romer called the strategy "plausible" but "challenging". This strategy is currently hampered by poor consumer confidence.

"A billion people is a lot of brainpower and there's a lot of potential for specialisation and supplying different kinds of services," Romer said. "[For] a country at the level of development of India or China, logically there's no reason why they have to be reliant on the rest of the world."

In getting to its current level of development, though, China relied heavily on protectionism, in part over concerns of foreign influence. Its technology giants that are today best positioned to take advantage of the AI boom - including industry leader Baidu, Tencent Holdings and Alibaba Group Holding, owner of the South China Morning Post - grew up without any significant competition from the likes of Google, Facebook and Amazon.com.

Some countries have tried to emulate this success, but Romer suggested that it is better to let investors find out where the best opportunities lie for a given market.

"I wouldn't recommend that these developing countries think first about what infant industries should we be protecting," he said. "This idea of the national champion I think is probably just a distraction."

There are now rising concerns about increasingly bifurcated technological spheres of influence, as the US escalates export restrictions on China for things like advanced semiconductors, and China works to counter those effects.

China's Belt and Road Initiative, consisting of costly infrastructure projects in mostly developing economies, has been one way Beijing has sought to bolster its soft power. But in Romer's view, making this kind of project work requires a more traditional kind of investment: real estate.

"I think Belt and Road is not going to prove to be self-financing unless they expand from ports and roads and rail into something like real estate development at these hub points that they're creating," he said.

The ideal scenario, according to the economist, is that the US and China compete to build up cities in developing markets, spurring growth based on merit.

"I hope they compete vigorously to be the country that's running the destinations that more people want to go to," Romer said. "It's very important for the US and China to figure out how to compete with each other in a way that actually makes the world better off rather than in a way that could really threaten everybody."

This article originally appeared in the South China Morning Post (SCMP), the most authoritative voice reporting on China and Asia for more than a century. For more SCMP stories, please explore the SCMP app or visit the SCMP's Facebook and Twitter pages. Copyright © 2024 South China Morning Post Publishers Ltd. All rights reserved.

Copyright (c) 2024. South China Morning Post Publishers Ltd. All rights reserved.