THE HIDDEN SIDE OF GENERATIVE ARTIFICIAL INTELLIGENCE: INVISIBLE LABOR, ENERGY CONSUMPTION, AND ETHICAL CHALLENGES IN THE CASE OF CHATGPT

Abstract

Generative artificial intelligence, represented by tools like ChatGPT, has revolutionized content production, communication, and task automation. However, little discussion surrounds the hidden impacts of its creation and maintenance. This article aims to problematize the invisible costs of generative AI, exploring three main axes: intensive energy consumption, dependence on invisible human labor, and the ethical implications of this technological model. Training language models requires large volumes of data and computing power, resulting in significant carbon emissions and energy consumption, which raises environmental concerns. Furthermore, behind the supposed autonomy of AI, there are thousands of workers, often in countries of the Global South, who perform labeling, filtering, and content moderation tasks under precarious working conditions, constituting so-called ghost work. Finally, the article discusses algorithmic opacity and the centralization of power in a few corporations, factors that hinder the democratic governance of technology. Based on scientific articles, technical reports, and investigative reporting, this study seeks to offer a critical and up-to-date analysis of the side effects of generative AI, proposing reflections on technological justice, transparency, and sustainability. By revealing what lies behind the intelligent interface, it hopes to contribute to an ethical and informed debate about the future of artificial intelligence.

DOI: https://doi.org/10.56238/sevened2025.030-012