Artificial intelligence is rapidly changing how people learn, work, and demonstrate what they can do. As AI systems take on tasks once associated with highly educated professionals, long-standing assumptions about education, skills, and economic value are coming under pressure. In a new article published on September 12, 2026 in ECNU Review of Education , Yilei Shao examines what these changes mean for the economics of education and asks a fundamental question: Does human capital remain an adequate framework for understanding education when intelligence is no longer scarce?
The paper begins with changes already visible in the labor markets. Major technology companies have reduced their reliance on traditional academic credentials, while AI systems can now provide expert knowledge and problem-solving capabilities on demand. According to the article, these developments challenge the assumption that educational credentials remain reliable indicators of competence and raise broader questions about the future value of formal education.
To explore the implications, the study revisits influential theories in education and economics associated with five Nobel Prize-winning scholars. It identifies five assumptions that have shaped modern education systems: linear returns to education, stable skill depreciation, individual ownership of capability, stable signaling through credentials, and largely positive educational externalities.
The article argues that AI is disrupting all five assumptions at once. First, returns to education are becoming less reliably linear. The labor market increasingly rewards people who can cross critical thresholds in human-AI collaboration rather than simply accumulate additional years of schooling. Second, skill are depreciating more rapidly, with some competencies becoming outdated within only a few years. Third, capability is increasingly distributed across human-machine systems rather than residing entirely within individuals. Fourth, educational credentials may become less reliable signals of competence when AI can perform many of the tasks associated with credentialed expertise. Finally, educational externalities are becoming more complex, as the knowledge and experience used to train AI systems can generate benefits captured by technology platforms rather than by those who contribute the underlying data and expertise.
In response, the paper proposes Intelligence Surplus as a new central concept for the economics of education. It defines Intelligence Surplus as a society's accumulated reserve of intelligent capability beyond its basic operational needs, capacities that can be mobilized for innovation, adaptation, and institutional development. Unlike human capital, which is primarily understood as an attribute of individuals, Intelligence Surplus is conceptualized as a societal and systemic resource.
Drawing on Pierre Bourdieu's theory of cultural capital, the author identifies three forms of Intelligence Surplus. The first is embodied, encompassing human capacities for collaborating with AI and navigating complex human–machine environments. The second is objectified, represented by technical systems such as AI models. The third is institutionalized, consisting of governance systems, rules, standards, and accountability frameworks that shape how intelligent technologies are used.
A central argument of the paper is that the key scarcity of the AI era is no longer intelligence itself. Instead, the emerging scarcity lies in what the author calls meaning-translation capacity: the ability to turn unique human experiences, judgments, and values into forms that can interact productively with AI systems. The paper introduces the concept of the First Compressor to describe individuals who translate irreplaceable human experiences into representations that broader intelligent systems can understand and process. According to the author, these capacities are difficult to automate because they depend on human relationships, care, interpretation, and judgment.
These ideas point to a broader rethinking of what education is for. Rather than primarily preparing workers with marketable skills, education is presented as the social mechanism through which societies create, sustain, and renew Intelligence Surplus. The author therefore proposes understanding education as a form of intelligent infrastructure, whose role is to support the creation, circulation, governance, and renewal of intelligent capabilities across society.
The paper outlines three interconnected levels of educational transformation. At the individual level, education should cultivate dual literacy, bringing technological literacy together with humanistic literacy. At the organizational level, educational institutions should create environments that support meaningful human-AI collaboration and lifelong learning. At the societal level, governments should build educational systems capable of functioning as intelligent infrastructure.
Looking beyond economic productivity, the author ultimately returns to a classical philosophical question: What is education for? Drawing on Hannah Arendt's distinction between labor, work, and action, the paper argues that education's deeper purpose is to cultivate people capable of exercising judgment, caring for others, creating meaning, and acting responsibly within society. As AI takes on more routine cognitive work, these distinctly human capacities become not less important, but more central to the purpose of education.
ECNU Review of Education
Commentary/editorial
Not applicable
Intelligence Surplus: Economics of Education After Human Capital
12-Sep-2026
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.