Hyperallergic editors call out AI-generated writing in the art world

Artificial Intelligence has found its way into museum press releases, exhibition texts, and artist statements, and editors at the independent arts publication, Hyperallergic, are growing increasingly frustrated. Their criticism has brought a familiar debate about AI into a particularly sensitive corner of the cultural sector, where the interpretation of art depends on knowledge, original ideas, and a true understanding of the artists behind the work.

In an opinion piece titled 'Dear Art World, We Can Tell When You're Using AI', Hyperallergic senior editor, Valentina Di Liscia, criticised the growing presence of generative writing in professional arts communication. She described receiving an apparently AI-written press release from a respected Manhattan institution, an example she used to illustrate how readily automated language has entered the industry's everyday communications.

Associate editor, Lisa Yin Zhang, echoed the criticism in a separate editorial, drawing attention to the repetitive sentence structures and exaggerated language associated with generative AI. Both editors expressed frustration with writing that appears polished but offers little meaningful insight into the art that it describes.

Their concerns don't just lie within stylistic preferences. Museums, galleries, and cultural organisations rely on written material to introduce artists, explain exhibitions, and establish historical context. A wall label can inform visitors and enhance their understanding of an unfamiliar work. A press release often provides the foundation for subsequent media coverage. When institutions rely on automatically generated descriptions, they risk circulating generic interpretations, factual inaccuracies, and language that fails to reflect the artist's intentions.

The criticism also raises questions about the value cultural institutions place on professional writers. Producing an informed exhibition text requires research, conversations with artists and curators, and careful editorial judgement. Generative AI can assemble convincing prose within seconds, but its ability to produce fluent sentences offers no guarantee of accuracy or critical understanding. The responsibility of what an institution publishes ultimately remains with the people whose names and reputations appear alongside it.

Di Liscia nevertheless acknowledges the difficult working conditions that have made AI attractive to many artists and creators. Independent practitioners frequently manage their own publicity, prepare funding applications, and write promotional material alongside their creative work. These responsibilities demand time and skills that artists rarely have time to develop, particularly when they have no gallery representation or administrative support.

However, for larger institutions, the expectations are different. Museums and foundations that have the resources to employ communication specialists, commission writers, and maintain editorial teams have no reason to rely on generative AI when there are jobs for that exact purpose. Di Liscia also recognises that institutional employees often work under considerable pressure, making adequate staffing and realistic workloads essential to addressing the problem.

Within arts administration, there are legitimate reasons for AI use, such as organising research materials, improving accessibility, and assisting with routine communications. Its adoption also questions where institutions should draw the line between technological assistance and the creation of original cultural commentary. Clear editorial policies, human fact-checking, and transparency about substitution AI involvement could help organisations establish those boundaries.

A main issue here is that identifying AI-generated writing is not always straightforward. Repetitive language, vague descriptions, and inflated claims have existed long before LLMs became widely available. Suspicion based on writing style alone cannot establish authorship, and institutions should be assessed on the accuracy, originality, and integrity of their published material.

This debate has arrived at a time when cultural organisations are already confronting many issues with AI, including the use of copyrighted creative work in training datasets, the displacement of creative labour, and the responsibilities of institutions that champion artistic expression.

Hyperallergic's intervention brings those concerns directly into the editorial process. Museums and galleries invest considerable resources in preserving, exhibiting, and interpreting creative work. The language they publish forms part of that responsibility, influencing how artists are represented and how audiences experience their ideas. As generative tools become increasingly imbedded within institutional workflows, the quality of that writing, and the human expertise behind it, deserves the same attention as the art it seeks to explain.