As a new-generation versatile technology, artificial intelligence is prompting economists to reconsider the fundamental principles underlying growth models, labor markets, and wealth distribution. Central to this discussion is a divide between optimism about technology, which foresees “exponential growth,” and labor economists, who caution about the possible “devaluation of human capital.”
Scholars are engaged in a theoretical debate on whether AI can enable the broader economy to break free from its long-standing pattern of linear growth, with differing perspectives rooted in microeconomic assumptions. Traditional semi-endogenous growth models indicate that although global investments in research and development have continued to rise, diminishing marginal returns have kept the annual growth rate of per capita gross domestic product in developed nations around 2 percent for an extended period.
The growth debate intensifies as AI’s reasoning and cognitive abilities reach certain thresholds. Some experts, like a University of Virginia economist, suggest that once AI surpasses these thresholds, it could transition from a passive support role to functioning as a “synthetic R&D workforce” capable of autonomous innovation, potentially propelling economic expansion beyond its current linear path and accelerating growth.
Conversely, Stanford economist Charles Jones introduces the “weak-link” model, arguing that the ultimate limit of economic expansion will be dictated by the most challenging tasks to automate, which serve as bottlenecks for further growth. In the labor market, AI is making significant inroads into cognitive and decision-making tasks for the first time, raising concerns that premium wages for cognitive specialists might decline.
Nobel laureate economist Daron Acemoglu proposes a “pro-worker AI” framework, differentiating between technologies that automate existing tasks and those that generate new ones. Automation of tasks that simply replicate skilled professionals’ tacit knowledge could lead to a reduction in labor’s income share. However, AI that complements human judgment can enhance the value of expert skills through collaboration.
For instance, in fields like precision engineering and medical diagnostics, AI can handle initial analyses of unstructured data, enabling professionals to concentrate on strategic reasoning and validation. Acemoglu emphasizes that the primary structural challenge in labor markets stems from excessive capital investment in purely automation-focused technologies.
Meanwhile, extremely low marginal costs may undermine traditional economic measures such as GDP. If AI significantly reduces the costs of producing software, professional services, and certain physical goods, money circulation within markets might decrease. This could lead to a misrepresentation of economic health; consumers could enjoy increased surplus and improved access to services, yet standard indicators might falsely suggest an economic slowdown or deflation.
Economists point out that current tax systems in many developed countries tend to impose higher payroll taxes on labor while offering incentives like accelerated depreciation for capital investments. This economic environment encourages firms to prioritize automation over human employment. They argue that addressing these distortions requires a combination of policy actions, including tax reforms, strengthened antitrust laws, and clearer rules around data ownership.