How AI Will Make Art Worse

Technological Change and Shifts in Artistic Practice: Historical Patterns and AI Developments

Many observers have expressed concern that artificial intelligence systems could reduce demand for work by human artists or alter established production methods in fields such as screenwriting and visual art. Examples include labor actions by Hollywood writers concerned about large language models and sales of works by robotic systems such as Ai-Da. Historical records, however, show repeated instances in which new tools prompted adjustments rather than elimination of prior practices. Technological developments have often redirected attention from technical execution toward choices about subject matter and intent. On this basis, some analysts anticipate that AI tools may encourage artists to emphasize personal experience and variation over uniformity of finish.

Photography provides one documented case. After its introduction in the nineteenth century, some painters expressed alarm that the medium would displace painting. The French artist Paul Delaroche reportedly stated in 1840 that painting had ended. In practice, painters responded by exploring approaches less suited to mechanical reproduction. Movements including Impressionism, Fauvism, and Cubism placed greater weight on subjective perception and simplified or fragmented forms. Painters asked what remained distinctive once mechanical capture of appearances became routine.

Contemporary artists face an analogous question. AI systems differ from earlier tools in versatility and speed of iteration. Their outputs can be adjusted through prompts without fixed physical constraints, and development trajectories remain subject to disagreement among researchers. One consistent observation is that human artists retain the capacity to accumulate lived experience that differs across individuals. Works produced under these conditions are therefore expected to incorporate more specific biographical references and personal idiosyncrasies. Group affiliations may give way to more individualized approaches or to collaborations that preserve distinguishable contributions.

Consistency of style can increase the ease with which automated systems replicate surface features. Systems currently generate images that match established studio aesthetics, such as those associated with Pixar or Studio Ghibli productions. Animators whose output varied across projects, including Richard Williams and Satoshi Kon, present greater technical challenges for replication. Observers therefore note that sustained variation and departure from prior work may become more common strategies.

These tendencies appear in existing online formats. Video essays on platforms such as YouTube have shifted in some cases from edited compilations with voice-over to direct-to-camera presentations with minimal post-production. Creators cite both the desire to signal human authorship and the opportunity to foreground individual mannerisms that differentiate one channel from others.

Current AI systems produce outputs that sometimes contain factual or stylistic inconsistencies. Continued improvement could reduce such occurrences. At that stage, systems might generate works that address thematic or emotional concerns without ongoing human guidance. Under those conditions, distinctions between human and automated production could rest more explicitly on process rather than on measurable competence in execution. Human artists might then explore approaches that retain visible traces of revision, uncertainty, or partial resolution.

Historical responses to new tools have sometimes involved deliberate departure from prevailing standards of finish. If automated systems achieve high consistency in both structure and surface, some human work may instead retain elements that record decision sequences or material limitations. Sketches, unrevised drafts, and provisional forms could receive greater attention as records of a particular maker’s activity. Errors or abrupt shifts might be retained rather than corrected, functioning as markers of authorship rather than defects.

The article’s author, Tim Brinkhof, presents these developments as one possible trajectory rather than an inevitable outcome. The argument draws on documented changes following the introduction of photography and on observable patterns in current digital production. It does not claim that all artists will adopt the same responses or that quality judgments can be settled in advance. Future outcomes will depend on choices made by artists, audiences, and institutions as tools continue to evolve.

The relationship between tools and artistic norms has never been fixed. Each new technology has altered the range of feasible actions and thereby changed which qualities receive emphasis. AI systems constitute one further instance of this pattern. Whether resulting works are evaluated more highly or less highly will depend on criteria applied by viewers and participants rather than on properties inherent to the technology itself.