The global scientific publishing industry has undergone a profound transformation over the past several decades, evolving from a system designed to disseminate knowledge into a multi-billion-dollar commercial enterprise that increasingly prioritizes profit over truth. This article argues that the commercialization of scientific journals and the rise of article processing charges have degraded peer review, incentivized predatory journals and paper mills, and undermined trust in the scientific record. Even prestigious venues are no longer immune to large-scale fraud, ghostwritten industry science, and retractions that continue to be cited as if nothing had happened. At the same time, a large fraction of genuine scientific work remains invisible, uncited, and effectively lost inside paywalled or low-visibility journals. We analyze these structural failures and describe how artificial intelligence can enable a new publishing paradigm: independent self-publication with DOI assignment, AI-assisted quality and integrity assessment, multidimensional trust scoring, and AI-driven cataloguing and cross-referencing of research. We propose an architecture in which AI and human evaluation are explicitly separated yet combined, ensuring that rigorous analysis is complemented by human capacity to recognize genuinely novel, paradigm-challenging work. This framework aims to disconnect the evaluation of scientific truth from the profit motives of commercial publishers and to inaugurate a new era in which scientific knowledge is openly accessible, dynamically evaluated, and globally discoverable.
Zakharenko et al. (Tue,) studied this question.