The Economic Impact of Generative Artificial Intelligence: A Review of Research from the Perspective of Task Content

Jingyi Shi · Financial economics research. · 29 Oct 2026

Where this sits in the debate

Will AI actually boost productivity and growth? · its finding: it depends (why studies differ)
Will AI destroy more jobs than it creates? · its finding: it depends (why studies differ)

Direction of the finding as read from the abstract by an AI model (AI-written).

Key takeaway AI-written

A review of research on generative AI's economic impact through the lens of task content, covering firm adoption, productivity, market structure, labour substitution and complementarity, and employment structure. It finds a basic consensus that generative AI works mainly by reshaping tasks within occupations, and sets out where and why studies still disagree.

The paper in three steps AI-written

Written by an AI agent from the paper's abstract, which is the primary source and is quoted below.

The question

What does task-based research say about generative AI's effects on industries and labour markets, and why do studies disagree?

The approach

A systematic review of studies on firm adoption, productivity, market structure, labour substitution and complementarity, and employment structure, organised by a common three-effect framework.

The findings

The literature broadly agrees that generative AI acts by reshaping tasks within occupations, but disagrees on total employment effects, how far productivity gains are realised, and differences across countries.

How to read it: review. A synthesis of existing studies — breadth over novelty; quality depends on what it includes and excludes.

Limitations and counterpoints AI-written

The reviewed research covers short time spans, has little evidence from developing economies and little on long-run dynamics.

Abstract

The rapid iterative development of generative artificial intelligence technology is fundamentally reshaping production methods and labor forms. Its ability to penetrate non-routine cognitive tasks breaks the impact boundary of traditional automation technologies, and also promotes task content analysis method to become the mainstream research paradigm in this field. This paper systematically sorts out the theoretical evolution and research progress of the economic impact of generative AI from the perspective of task content. From the two dimensions of industrial development and labor market, it integrates the research results of core topics such as enterprise adoption, productivity effect, market structure, labor complementarity and substitution, and changes in employment structure. It focuses on clarifying the sources of divergence and applicable boundaries of different research conclusions. Studies show that existing literature has formed a basic consensus that generative AI exerts impact mainly through the reconstruction of intra-occupational tasks. However, there are still significant divergences in the total employment effect, the implementation degree of productivity gains, and the mechanism of cross-country heterogeneity. The main limitations of current research in this field include the short research time span, lack of evidence from developing economies, and insufficient research on long-term dynamic mechanisms. The contribution of this paper is to restore the seemingly divergent literature conclusions to differences in methodological premises with a unified three-effect coordinate, and put forward corresponding testable designs for each type of research gap. Future research needs to be further deepened around the two core dimensions of industry and labor market, improve localized measurement systems, and expand long-term dynamic analysis.

About this publication

Academic journal. Published in an academic journal — typically peer-reviewed before publication, though indexing alone doesn't guarantee review depth.

Published: 2026 (recently)
Citations: 0 (OpenAlex)