Artificial intelligence (AI) is reshaping the education sector, yet the economic dimension of this transformation remains largely underexplored. While much of the current discourse emphasizes the pedagogical, ethical, and societal implications of AI, few studies have sought to quantify its potential to improve efficiency and reduce costs within education systems. This paper addresses that gap by proposing a structured analytical framework to estimate the economic savings AI could generate when integrated thoughtfully into educational processes.

We develop and contrast three economic scenarios:
(A) AI ignored, in which current structures remain unchanged;
(B) human substitution, in which AI replaces significant portions of educational labour; and
(C) augmented human action, a cobotic approach where AI supports and enhances human capabilities.

The third scenario aligns with a vision of human–AI collaboration that values professional judgement, equity, and pedagogical intent (Huang et al., 2019; Adams & Thompson, 2025).

Building on the concept of cobotics—the coordinated distribution of tasks between AI and humans—we explore areas where AI may generate measurable efficiency gains: automation of administrative and assessment tasks, personalization of learning pathways, predictive resource allocation, and data-informed decision-making at the institutional level. By cross-referencing data from public budgets (Quebec and Canada), international projections (PwC, UNESCO, OECD), meta-analyses of AI uses in education (Zawacki-Richter et al., 2019), and case studies of institutional practices to propose exploratory estimates: in some contexts, AI could reduce operational costs without compromising educational quality.

However, beyond quantification, the central question remains: what should be done with the gains generated? Following concerns raised by Brandusescu (2021) and Duhaime (2022) about the capture of public value by private actors, we argue that economic gains must be redistributed equitably.

We propose four guiding principles:
(1) reinvestment in human-centered educational services, particularly those supporting vulnerable learners;
(2) funding of teacher and leadership training for AI acculturation (Gaudron, 2024);
(3) support for public-interest educational research and open-source solutions; and
(4) governance frameworks ensuring democratic accountability and ethical oversight of algorithmic tools (UNESCO, 2021; European Commission, 2019).

By combining prospective modelling with policy analysis, this paper contributes to a necessary reframing of AI not just as a technological or pedagogical innovation, but as a public infrastructure whose economic implications must be actively governed. At a time when education budgets are under pressure and the promises of AI are highly mediatized, this work provides a foundation for evaluating the conditions under which AI can truly serve the public good in education.

Keywords: Artificial intelligence, education, cobotics, economic modelling, efficiency, redistribution, public policy.