Source-linked AI summary
Can Computers Create Art?
Aaron Hertzmann
TL;DR
This essay examines whether AI systems can create art and be credited as artists, placing current debates alongside earlier technological changes in art. It theorizes that art is a social activity and concludes that software systems cannot currently be called artists, while identifying possible challenges and risks surrounding that boundary.
Problem
Debate persists over whether computers are merely artists’ tools or can themselves be artists, amid fears that AI may replace artists and claims that algorithms could be artists.
Method
The essay surveys photography and computer animation, discusses current AI art tools, and theorizes prerequisites for being an artist through art as social behavior and alternative attribute hypotheses.
Results
The essay concludes that, in current understanding, software systems cannot be called artists because art is a social activity, while AI tools are likely to become recognized as artists’ tools.
Takeaways & Limitations
Calling shallow AI systems artists may mislead people about their emotions, ethical weight, competence, and the nature of art.
Takeaways & Limitations
Meaningful artistic growth and meaningful responsiveness to culture, experiences, events, feedback, and environment remain difficult for current AI systems and may require enormous advances or social AI.
Abstract
from arXiv · showhide
This essay discusses whether computers, using Artificial Intelligence (AI), could create art. First, the history of technologies that automated aspects of art is surveyed, including photography and animation. In each case, there were initial fears and denial of the technology, followed by a blossoming of new creative and professional opportunities for artists. The current hype and reality of Artificial Intelligence (AI) tools for art making is then discussed, together with predictions about how AI tools will be used. It is then speculated about whether it could ever happen that AI systems could be credited with authorship of artwork. It is theorized that art is something created by social agents, and so computers cannot be credited with authorship of art in our current understanding. A few ways that this could change are also hypothesized.
1 Introduction
The essay frames AI art as a question about authorship and social understanding, not merely technical capability. It argues that new technologies can expand artistic tools and opportunities while computers remain tools rather than accepted artists.
- AI advances and media hype have intensified fears that image- and video-making systems may empower users while displacing artists.
- The essay distinguishes technical capability from whether software will be widely considered an artist or credited as an artwork’s author.
- Historical automation of photography and film initially raised fears of artist displacement but created new opportunities and invigorated traditional media.
- The author argues that AI technologies will provide new tools and modes of expression, transform art and culture, and create new job opportunities.
- Existing computer-generated art is credited to human artists because human invention, software development, tweaking, and direct control remain central to its production.
2 History and Current Practice
Historical examples show that technologies automating image and film creation initially threatened traditional artists but ultimately expanded artistic tools, styles, and professional opportunities. Computer art and animation likewise depend on human-defined procedures and collaboration, while current AI systems remain tools rather than accepted artistic authors.
- Historical perspective: Photography, film, and computer animation illustrate how technologies feared as replacements for artists created new opportunities and roles instead.The essay uses these historical cases to frame technology as expanding rather than simply displacing artistic practice.
- 2.1 How photography became an artform: Photography’s displacement of older image-making techniques pushed painters beyond hyper-realistic depiction and helped stimulate modern art.The essay links photographic influence to abstraction, artistic experimentation, and renewed vitality in painting.
- 2.1 How photography became an artform: As photography became more controllable and widespread, it became an accepted artistic tool and made image-making accessible to non-experts and hobbyists.The essay predicts a similar trajectory for controversial AI tools.
- 2.3 3D computer animation: a collaboration: Computer animation developed through collaboration between technical staff and artists, with new technologies answering artists’ needs while inspiring further artistic experimentation.Pixar’s development is presented as an interaction in which art challenges technology and technology inspires art.
- 2.4 Procedural artwork: Procedural art and painterly rendering automate aspects of production through rules, algorithms, or learned image transformations, but the human-defined procedure remains central.Examples include AARON, interactive and automatic painterly systems, Image Analogies, Neural Style Transfer, and DeepDreams.
- 2.5 State of the art in computer science research: The essay argues that current AI-based art systems are tools comparable to brushes and paint, with a human still behind the artwork.It distinguishes surprising outputs from genuine artificial intelligence or independent authorship.
3 How technology changes art
The essay argues that new technologies, including AI, generally become tools that expand artistic expression rather than simply replace artists. Historical examples suggest technology can make art more accessible, sustain cultural vitality, and generate new styles, while specific future effects remain difficult to predict.
- New technologies become tools for artists, invigorating and changing art and culture rather than simply replacing artists.
- 3.1 Algorithms are artists’ tools: Current AI systems remain human-directed tools: artists select inputs, adjust settings, modify code, and choose desirable outputs.The essay describes neural style transfer and DeepDream as dependent on human invention, experimentation, and selection.
- 3.2 New technology helps art stay vital: Technological innovation can keep art vibrant by enabling new styles, sounds, and forms that challenge older conventions.The essay uses oil paint, synthesizers, photography, and cinema to illustrate how new media alter artistic possibilities.
- 3.3 New technology makes art more accessible: New artistic tools also broaden participation by making image-making and other creative practices more accessible to non-experts and hobbyists.Photography and modern computing are presented as examples of technologies becoming easier, faster, and more widely available.
- 3.4 New AI will be new tools for artists: AI is expected to provide simplified tools and amplify imagination, although some mechanical artistic tasks may fade and specific future developments are hard to forecast.The essay distinguishes robust general trends from unpredictable particular inventions and artistic uses.
4 What is an Artist?
The essay argues that art is fundamentally a social interaction between social agents, so technical skill, novelty, or autonomous output alone does not make a computer an artist. Authorship depends on whether viewers infer a socially situated being behind the work.
- 4.1 Art is social: Art is primarily communication and display between people, with personal expression functioning as an act of communication.
- 4.1 Art is social: Art is hypothesized to be an interaction between social agents—entities accorded personhood, empathy, and ethical consideration.
- 4.1 Art is social: Non-social benefits such as skill practice, pleasure, and meditation are secondary to art’s social functions, even when art-making occurs alone.
- 4.2 Skill: Technical proficiency is insufficient for authorship: computers can generate highly realistic or abstract imagery, while unskilled humans can still make art.
- 4.3 Intelligence and Consciousness: Viewers may judge whether a computer is an artist by seeking evidence that it is intelligent, conscious, feeling, and therefore a social being.
- 4.4 An Intent Machine: An Intent Machine would likely be attributed to its system-builder because its procedure resembles procedural art, even if its resulting work were good.
- 4.5 Creativity, Growth, Responsiveness, and so on: Surprising or complex output is too weak a creativity criterion: the Mandelbrot set produces dazzling images without its iteration equation being considered creative or artistic.
- 4.5 Creativity, Growth, Responsiveness, and so on: Current procedural systems may lose novelty because they do not grow or evolve meaningfully over time like human artists.
5 Will an AI ever be an artist?
The essay considers whether software could ever receive artistic authorship, distinguishing human-level AI from shallower systems perceived as social agents. It concludes that current non-social software is unlikely to qualify, while warning that labeling it an artist can mislead audiences and obscure human authorship.
- 5 Will an AI ever be an artist?: Current algorithmic art is credited to human authors, but the essay asks whether software could ever be widely recognized as an artwork’s author.
- Human-Level AI: Human-level AI would be able to create art, but this remains a speculative scenario whose possibility, implementation, and social consequences are unknown.
- Human-Level AI: The more focused question is whether computer-authored art could exist without human-level intelligence or consciousness.
- Social AI: AI could receive authorship if people viewed it as a social agent performing communication or sharing through art.
- Social AI: Shallow AI might be perceived as social without being truly intelligent, as illustrated by people attributing human-like emotions to ELIZA.
- Non-social AI: Non-social algorithms might be promoted as artists, but acceptance as true artists would likely require plausible social or conscious attributes.
- Dangers: Calling shallow AI an artist risks misleading people into attributing emotions, feelings, ethical weight, and excessive competence to it.
- Dangers: Assigning authorship to software can deprive designers of credit, although current practice recognizes their skill and effort in engineering good outputs.
Conclusion
The conclusion rejects calling current software an artist because art is treated as a social activity, while emphasizing technology’s positive contribution to artistic innovation. AI is presented as another tool that may help art evolve.
- Conclusion: No software system in the essay’s current understanding qualifies as an artist because art is a social activity.
- Conclusion: Technology contributes to art by giving artists increasingly powerful tools through which new art forms can develop.
- Conclusion: Art remains vital through continual innovation, and technology is described as one of the main engines of that innovation, including emerging AI experiments.