arXiv cs.AI: Liapis and Panagopoulos argue that AI outputs can reshape shared meanings

arXiv cs.AI: Liapis and Panagopoulos argue that AI outputs can reshape shared meanings

In their arXiv abstract, Charalampos M. Liapis and Epameinondas Panagopoulos argue that language-generating AI outputs can return to the human and institutional practices that shaped their training material. They call this process recursive semantic normalization: machine-generated formulations, when repeatedly taken up, evaluated, reused, and retained by institutions, can help reproduce or transform semantic regularities. Their concept of technohybridism treats human practices, computational generation, and institutional uptake as a coupled system, rather than treating model outputs alone as the unit of analysis. The authors distinguish generation, selection, uptake, and persistence, and say their model separates changes attributable to machines from changes that would have happened anyway. They identify possible social-epistemic consequences: provenance laundering that weakens epistemic vigilance, erosion of cognitive diversity, and narrower shared hermeneutical resources. Their concern is not that machine outputs acquire authority: they explicitly say their account attributes no normative authority to those outputs. Rather, recurrent human and institutional uptake may change the field in which people form and share meanings. The abstract presents an analytical account and an evidential ladder for empirical research; it does not report findings establishing how often these [

Publication: 39eb8108-840d-4130-8adb-c294cd7e54f2

attribution — pass: The draft correctly attributes the arXiv paper to Charalampos M. Liapis and Epameinondas Panagopoulos and keeps the claims as the authors' arguments. Labeling it 'arXiv cs.AI' matches the publisher field, although the primary listing is cs.CY.

Authors:Charalampos M. Liapis, Epameinondas Panagopoulos

summary — uncertain: The summary is cut off mid-sentence ('how often these ['), so the closing limitation is incomplete; repair by finishing that sentence. Central terms are also left unexplained for non-specialists: semantic regularities, provenance laundering, epistemic vigilance, cognitive diversity and hermeneutical resources. These should have term-library tooltips or entries. The draft also omits the authors' analysis of uptake as routinized, institutionally diffused and socially stratified, and the conditions under which machine-derived regularities become norms. Restoring one clause on that point would help. The question is substantive and follows from the evidence.

we analyse uptake as routinized, institutionally diffused, and socially stratified practice and state conditions under which machine-derived regularities become norms

date — pass: originalPublished is 2026-11-05, which matches the v1 submission date. The later feed timestamp was correctly not used.

[Submitted on 5 Nov 2026]

support — pass: Every claim is backed by a verbatim abstract excerpt. The draft also correctly notes that no empirical findings are reported. Calling the consequences 'possible' slightly softens the authors' direct 'we identify' framing. This is a minor fidelity point that does not add unsupported material; dropping 'possible' or writing 'consequences the authors identify' would preserve their force.

Fourth, we identify social-epistemic consequences: the laundering of provenance that weakens epistemic vigilance

media — pass: The selected media item is the verified direct arXiv abstract page for this paper, which is the most direct access point to the source.