Harnessing Comparative Advantage: The Key to AI-Driven Prosperity
Rapid advancements in artificial intelligence (AI) raise concerns about unemployment and inequality, but economic theory offers a more optimistic view. The concept of comparative advantage suggests that while AI will automate some jobs, many roles will remain intact or evolve, enhancing productivity and wages, especially for those who adapt and collaborate with this new technology. However, growing wage inequality remains a concern, and effective policies will be necessary to mitigate these disparities and ensure AI promotes global equity. Ultimately, while AI presents challenges, it also offers significant opportunities for innovation and economic growth, making the enhancement of human capital through education and training crucial in this evolving landscape.
Rapid advancements in artificial intelligence (AI) and machine learning have the potential to reshape our world, but what will this change look like? Will AI lead to high levels of unemployment, poverty, and inequality? Luckily, economic theory suggests we may not need to be overly concerned; rather, we can be hopeful for a future full of “plentiful, high-paying jobs” (Smith, 4).
The key economic concept that holds out hope for a prosperous future is that of comparative advantage. As noted by journalist Noah Smith, no matter how efficient and productive AI gets, it will still be dependent on two major constraints: energy and computing power. There are always opportunity costs, and AI’s intense energy requirements and limitations on the development of computing capabilities are significant. For instance, Ontario’s demand for electricity “is set to soar by 75 percent in the next couple of decades,” largely driven by increased energy needs for AI development (CBC). As countries work to transition to cleaner energy sources, the costs of energy may skyrocket in the short run, potentially stalling some of AI's rapid growth.
Continuous improvement in AI is not guaranteed. Diminishing returns on machine learning and constraints on hardware production mean that AI may have limited short-term growth potential (Chau). Rock’s law suggests that that says that “the cost of a semiconductor [fabrication plants] doubles every four years” making continued developments in key hardware extremely costly (Smith, 10).
Given these constraints, AI will be primarily allocated to the most productive uses—likely complex tasks and computations that current capabilities cannot perform efficiently if at all. Because of this, quite a few existing jobs will remain relatively unaffected by the AI revolution, and as argued later, those who work them may find themselves better off. Another significant portion of the workforce may experience considerable changes to their roles, but in ways that enhance productivity and wages through collaboration with new technologies. The opportunity cost of AI is too high, for economy-wide automation to be efficient or plausible. Diminishing returns in AI productivity will make skilled human labour a necessary complement.
Jensen Huang, CEO of Nvidia, recently stated that the future of coding is in English, and AI is to thank (Forbes). From one perspective, the radical changes brought by artificial intelligence may make old skills obsolete; in this example, even those who work closely with AI may find their jobs changing significantly. But is this a cause for concern? Huang envisions AI as a tool for “closing the technology divide,” where coding languages are replaced by more accessible prompts that anyone can use (Forbes). AI can increase access to new technology and reduce barriers to entry into more productive industries, making previously inaccessible information available and existing processes more efficient and cost-effective. Rather than become obsolete, workers will find themselves with new tools and new, more productive, applications.
Imagine three types of jobs that will emerge in a new age with AI: jobs fully automated by new technologies, jobs that require human-AI complementarity, and jobs that remain entirely human-operated. In this new world, workers in the second category—those who complement AI—are likely to experience significant benefits. As technology and capital become more productive, so will the labour that works alongside it. Those who are able to integrate AI into new or existing processes will see real wages increase alongside productivity. Labour that is mobile and adaptable will reap the rewards of this new era significantly more than those that do not. Only tasks that are beyond current capabilities or that constrain more efficient allocations of human and physical capital will be fully automated.
There are many occupations that AI will neither automate nor complement, at least directly. As previously discussed, opportunity costs will limit what applications it will be used in no matter whether it is better at doing the task. Lower productivity jobs and those not easily replicated by technology will be the last to see major changes to their industry. For example, it is not worth using our limited stock of computational power to replace baristas when new technologies can unlock completely new and profitable domains elsewhere; it is not efficient to completely replace labour-intensive or low-remuneration industries when the opportunity costs of doing so are so high. The comparative advantage approach stipulates that as other aspects of the economy change and become more efficient, the real wages of workers in the human-operated jobs rise too. Everything else becomes cheaper with no substantial change in the demand for their services. It is hard to see the world demanding less coffee.
Yet, the integration of AI into the workforce also raises concerns about growing wage inequality. Research suggests that technology can exacerbate income and wealth disparities as it has focused productivity growth in service sectors (Van Reenan, 738; Lawrence and Slaughter, 209). While the real wages of both automated, complementary, and human-operated jobs all rise, history suggests that the rise may be unequal. However, with the right policies, AI could help mitigate these issues, leveling the playing field globally if correctly implemented and governed. How governments respond in the next few years of technological development will be a significant determinant of whether AI brings prosperity or peril. Indeed, being too heavy-handed might stifle its potential; however, insufficient protection against the risks of this new technology may greatly exacerbate cyber security concerns, compromise intellectual property rights, and increase the income gap.
In conclusion, while the rise of AI presents challenges, it also offers remarkable opportunities. By leveraging the potential of this technology through a framework of comparative advantage, we can envision a future not defined by unemployment and inequality, but by prosperity and innovation. The key to this future lies in enhancing human capital through education and training; mobile labour and innovation are key complements to continued technological growth and can help deal with the growing income gap. Worryingly, a look back into recent times suggests that we struggle in dealing with the kind of transnational cooperation issues that a future with AI entails. The major question will be how world governments go about making policies around it. The current political debate should center around managing a new stressor to the already dangerous energy crises, how to regulate against the potential safety and security risks, and securing access to these new technologies to reap their full economic potential on a global scale.
References
Brian Chau, "Diminishing Returns in Machine Learning: Hardware Development and the Physical Frontier," From the New World, 2022, Link.
John Van Reenen, "Wage Inequality, Technology and Trade: 21st Century Evidence," Labour Economics, 2023, Link.
Lawrence, Robert Z., and Matthew J. Slaughter, "International Trade and American Wages in the 1980s: Giant Sucking Sound or Small Hiccup?" Brookings Institution Press, 1993, Link.
Noah Smith, "Plentiful, High-Paying Jobs in the Age of AI," 2023, Link.
Tim Bajarin, "Nvidia’s CEO on the Democratization of Coding," Forbes, March 20, 2024, Link.
Allison Jones, "Ontario Electricity Demand Outlook," CBC, 2023, Link.