AI Artistry encompasses the visual culture produced through generative artificial intelligence tools, along with the debates over authorship, copyright, and creative labor that accompanied the rapid adoption of these technologies.

Origins and Early Development

DeepDream (2015) Google engineers open-sourced DeepDream in July 2015, a convolutional neural network visualization tool that produced hallucinogenic, pattern-heavy imagery. The psychedelic outputs became widely shared online, marking the first mainstream encounter with machine learning-generated visual art.

GANs and the Art Market (2018) In October 2018, “Portrait of Edmond de Belamy,” created by Paris collective Obvious using a Generative Adversarial Network trained on 15,000 historical portraits, sold at Christie’s for 432,500againsta432,500 against a 10,000 estimate. The underlying algorithm was based on open-source code by artist Robbie Barrat, sparking early debates about credit and originality in AI-assisted creation.

Text-to-Image Tools (2022) Three major text-to-image platforms launched within months of each other in 2022: DALL-E 2 (April), Midjourney (open beta July), and Stable Diffusion (August, open-source). These tools allowed users to generate detailed images from written prompts, dramatically lowering the barrier to visual creation.

The US Copyright Office issued guidance in March 2023 stating that copyright protects only human-authored elements of works, and that purely AI-generated content is not copyrightable. This principle was tested in several cases:

  • Zarya of the Dawn (2023): A graphic novel with AI-generated illustrations received partial copyright protection for its text and arrangement, but not for individual AI-generated images.
  • Théâtre D’opéra Spatial (2023): Jason Allen’s Colorado State Fair-winning artwork, created using Midjourney, was denied copyright registration.
  • In January 2023, artists filed a class-action lawsuit (Andersen v. Stability AI) against Stability AI, Midjourney, and DeviantArt, alleging copyright infringement via training datasets containing billions of scraped images.

Art Community Response

Competition Controversy (2022) In September 2022, Jason Allen won first place in the digital art category at the Colorado State Fair with an image created using Midjourney. The win generated widespread discussion among artists about whether AI-assisted entries should be permitted in fine art competitions.

Platform Protests (2022) When AI-generated images began appearing on ArtStation’s trending pages in late 2022, illustrators organized protests using “No To AI-Generated Images” graphics. DeviantArt launched its own AI generator trained on user artwork without explicit consent, prompting backlash that led the platform to opt all user artwork out of AI training by default.

Counter-Tools In 2023, researchers at the University of Chicago released Glaze, a tool that applies imperceptible perturbations to artwork to disrupt AI style mimicry, followed by Nightshade in 2024, which alters images so AI models misinterpret them during training.

Political Dimensions

AI image generation tools intersected with political communication in several ways:

  • In April 2023, the Republican National Committee released an entirely AI-generated attack ad depicting hypothetical scenarios.
  • In January 2024, a synthetic voice call impersonating President Biden told New Hampshire voters not to participate in the Democratic primary, leading to the first FCC enforcement action against political voice cloning.
  • California passed the Defending Democracy from Deepfake Deception Act in 2024, requiring platforms to label AI-generated political content near elections.

Visual Characteristics

The aesthetic evolved across technological generations. DeepDream-era outputs featured fractal-like patterns with recurring organic motifs. GAN-era imagery tended toward painterly, slightly blurred compositions. Diffusion model outputs from 2022 onward ranged from photorealistic to hyper-stylized, with common artifacts including malformed hands and inconsistent text. A recognizable “AI aesthetic” emerged, characterized by heightened saturation, smoothness, and an uncanny quality distinct from both photography and traditional digital art.