The real world does not change as quickly as AI changes
Real world, physical constraints pose limits on the rate at which A(G)I will change the physical world
Parts of this essay were written with help from ChatGPT’s new Canvas tool.
Introduction
Dario Amodei's recent essay, Machines of Loving Grace, outlines a vision for artificial general intelligence (AGI) that aims to transform the world for the better. Amodei believes that rapid scientific advances driven by AGI could lead to unprecedented global economic growth, especially in developing nations. While this outlook is ambitious and paints a hopeful picture of technology's transformative potential, critics like Tanner Greer offer a more nuanced critique that exposes some of the contradictions and practical challenges in Amodei's vision1.
Greer's criticism centers around two main points. First, he highlights the tension between keeping AI under the control of a liberal-democratic coalition while promoting AGI-aided governance in the developing world. Second, he questions whether compressing a century's worth of technological advances into a few years can realistically lead to similar economic gains, given the infrastructure and political challenges on the ground. These critiques demonstrate that the rate at which AI improves does not augur a similar rate of change in the physical world.
The Dilemma of Technology Transfer
Amodei argues for a coalition liberal-democratic Western governments2 to guide AGI development. We want powerful technology to stay in safe hands, and not be usurped by authoritarians. At the same time, Amodei suggests that AGI will guide industrial and economic policies in the developing world, potentially accelerating growth and contributing to global prosperity. However, Greer points out that these two goals may be fundamentally at odds with each other.
Keeping AGI within a liberal framework while deploying it across developing nations implies a significant degree of technology transfer. Historically, such transfers have often led to unintended consequences, including losing control over how the technology is used. Once advanced AI is introduced to local governments, the incentives to maintain the liberal-democratic values of its creators will weaken, particularly in countries with authoritarian governance or fragile institutions.
Moreover, the geopolitical dynamics of technology sharing cannot be ignored. Countries receiving AGI support may use it for purposes like surveillance or political oppression, which are incongruent with Western ideals. Implicit in Amodei’s argument is the idea that developing world governments would prioritize global prosperity over national interest. History has shown this to be overly optimistic.
Trust is also a major factor in this dilemma. Developing nations may be skeptical of the intentions behind AGI assistance from Western powers. It would be seen as a form of neo-colonialism or an attempt to exert undue influence. This distrust will hinder cooperation and lead to scenarios where developing countries’ governments resist AGI deployment or misuse it. Successful technology transfer involves not only the provision of AGI tools but also the creation of genuine partnerships, respect for sovereignty, and a deep understanding of local political dynamics. And, again, the historical record of failures suggests that optimism is not really warranted.
Technological Growth vs. Economic Development
Another key aspect of Greer's critique concerns the relationship between scientific advances and economic development. Amodei suggests that condensing a century of technological progress into a few years will lead to an explosion in global GDP, particularly in developing nations. Greer challenges this assumption by drawing on China's development as an example. He points out that much of China's economic growth in recent decades did not come from cutting-edge technology but from incremental advances in physical infrastructure.
China's rapid growth was driven by transitioning from carts to cars, from mud huts to skyscrapers, and from candlelight to electricity—all changes that required physical resources, labor, and long-term planning. Even if AGI provides advanced knowledge to accelerate industrialization, the physical, logistical, and political challenges of transforming infrastructure remain significant barriers. Knowledge doesn't build roads, upgrade plumbing, or replace makeshift shelters. These projects require physical labor, coordination, and substantial capital investment. Even if humanoid robots become widely available, there are still practical, political, regulatory, and local hurdles to overcome.
Moreover, economic development is not just about technological progress. It also involves societal changes, cultural shifts, and workforce skill development. Many developing countries face challenges like lack of education, poor healthcare, and inadequate governance, all of which can impede the rapid deployment of advanced technologies. For AGI to be transformative, it must be accompanied by investments in human capital, social infrastructure, and effective institutions that can support the long-term adoption and scaling of technological solutions.
Greer's example of China's campaign to install proper toilets in rural areas highlights the gap between technological advances and practical outcomes. Despite the availability of modern sanitation technology, the project faced major obstacles, such as geographic isolation, supply chain challenges, and local resistance. This example shows that even seemingly straightforward improvements require a coordinated effort that cannot simply be accelerated by better technology.
Uneven Development: A Persistent Reality
A deeper issue Greer raises is the phenomenon of uneven development. Within China itself, there are staggering contrasts—cities like Shanghai look like they belong in the future, while many rural areas still struggle with basic infrastructure. Despite China's economic success, vast inequalities exist between urban and rural regions. Economic growth is often benefits certain regions or sectors at the expense of others.
The Real Limits of AGI as a Global Panacea
The challenges Greer raises suggest that AGI, no matter how advanced, cannot bypass the material and logistical realities that underpin economic development. Technological advances can provide a blueprint, but they cannot by themselves supply the labor, physical materials, or governance structures needed to bring those blueprints to life. Moreover, assuming that AGI can effectively guide industrial policy in developing nations without risking geopolitical or authoritarian misuse is a significant oversight.
Amodei's vision reflects optimism that technology alone can solve problems deeply embedded in the socio-political and economic fabric of nations. This optimism overlooks the fact that development is not just about having the right answers; it is about navigating complex realities of implementation, which often involve conflicting interests, resource limitations, and human factors that cannot simply be engineered away.
Conclusion
Dario Amodei's essay paints a hopeful picture of AGI as a force for global good, but T. Greer's critique serves as an important reminder of the limitations in such visions. Technological solutions, no matter how advanced, cannot operate in a vacuum. The challenges of infrastructure, governance, political power dynamics, and uneven development are substantial hurdles that AGI alone cannot overcome.
Greer’s Twitter thread stands on its own; this essay is my attempt to flesh out the arguments he’s making.
Amodei doesn’t use the term but he seems to be arguing for WEIRD people to govern AI’s global rollout.