A newly published theoretical paper argues that generative artificial intelligence may be creating a novel form of informational inequality — one defined not by who can access technology, but by who critically evaluates what it produces.

The preprint, posted to arXiv under the title 'Generative Artificial Intelligence and the Knowledge Gap: Toward a New Form of Informational Inequality' (arXiv:2603.24335), extends decades of communications research into what scholars call the 'knowledge gap hypothesis.' First articulated in the 1970s, that hypothesis holds that the spread of new information technologies tends to widen, rather than close, social divides — a finding later reinforced by digital divide research focused on unequal internet access and digital literacy.

The paper's authors argue that generative AI demands a fresh theoretical lens. As access to AI tools becomes increasingly widespread — with chatbots and AI assistants freely available to broad swaths of the public — the old framework of 'haves versus have-nots' defined by access alone no longer captures the full picture.

Instead, the researchers propose that the critical fault line is now critical evaluation. Their central assumption: individuals with higher levels of education are more likely to question, contextualize, and scrutinize AI-generated outputs, while those with lower levels of education may accept such outputs more directly and uncritically.

This distinction carries significant implications. AI systems are known to produce confident-sounding but factually incorrect responses, reflect embedded biases, and omit important context. If less-educated users disproportionately rely on AI outputs without interrogating them, the result could be a new stratification of knowledge quality rather than mere knowledge access.

The paper is explicitly conceptual and presents no empirical findings. The authors acknowledge this limitation directly, framing the work as a theoretical scaffold intended to guide future research on the intersection of education, AI use, and knowledge inequality.

The contribution fits within a growing body of scholarship examining AI's societal implications beyond technical performance. Researchers in media studies, sociology, and communication have increasingly flagged that democratized access to powerful tools does not automatically translate to democratized benefit — a pattern observed with earlier information technologies from television to the early internet.

Whether empirical studies will confirm the paper's assumptions remains to be seen, but the framework offers researchers a structured starting point for investigating one of the more understudied risks of the generative AI era.

Source: arXiv:2603.24335 — https://arxiv.org/abs/2603.24335