Artificial Intelligence, Economic Inequality, and Labour Market Polarization: A Theoretical Perspective on the Future of Work

Abstract

Artificial intelligence is changing how work is organised, paid, and valued. This paper studies the link between artificial intelligence, economic inequality, and labour market polarization. It uses a theoretical method supported by public studies and published experiments. The main argument is that artificial intelligence does not affect all workers equally. It can raise output and reduce routine tasks for some workers. It can also weaken wages and job security for others. The effect depends on tasks, skills, institutions, and access to digital tools. Evidence from labour economics shows that earlier computerisation reduced many routine middle-skill jobs. Recent generative artificial intelligence reaches more cognitive and white-collar tasks. Experiments show strong productivity gains in writing, coding, consulting, and customer support. Yet these gains are uneven across occupations and worker groups. The paper explains this through task-based theory, skill-biased change, capital ownership, and institutional power. It also reviews public datasets from the World Bank, ILO, OECD, IMF, and Stanford AI Index. The paper finds that artificial intelligence can either reduce or increase inequality. It reduces inequality when it supports low-skill workers and spreads knowledge widely. It increases inequality when profits, data, and control stay concentrated. The future of work therefore depends on policy, training, bargaining power, and fair access. The paper ends with policy steps for inclusive adoption and social protection.