Module 07 - Creating Generative Art

Generative Art

The art world is currently reckoning with generative AI. While some individuals consider AI to replace human creators altogether, many others think of AI as an additional tool which aids in creation. There are many concerns about the training and/or input data used to power these models and whether creators’ consent is involved.

Table of Contents

  1. Generative Art
  2. Over 150 Generative Art Tools
  3. Potential Uses Cases in Science
  4. DALL-E vs. Adobe Firefly
    1. Prompt Given
    2. DALL-E via Bing Create
    3. Adobe Firefly
  5. Don’t Expect Too Much
  6. AI Artists to Follow
  7. Discourse on AI-Generated Art

Over 150 Generative Art Tools

The OpenTools website has curated a list of over 150 generative art tools which are currently available. Each may have different pricing models or availability.

Potential Uses Cases in Science

  • Personalized Lab Logos
  • Graphical Abstracts for Research Papers
  • Brainstorming Potential Journal Covers
  • Conceptual Icons for Slideshows
  • Badges for Github Repositories
  • Stickers to Advertise Your Projects
  • What are other ways generative art could benefit immunology research?

DALL-E vs. Adobe Firefly

Below are two generative art apps accessible via any internet browser. Here, we examine the differences when each software is given the same prompt to generate a logo for AI4Immuno:

Prompt Given

Art prompt

Alt Text: Image with light blue background with the following text printed, Cute cartoon logo with immune cells, bacteria, pathogens, and sparkles floating around a laptop showing the text: “AI 4 IMMUNO” (Source: AI4Immuno)

DALL-E via Bing Create

bing.com/images/create This option is currently free to use and appears to have a token limit, which “boosts” the speed of image generation for each query.

Adobe Firefly

firefly.adobe.com/ This option requires having a subscription to Adobe Creative Cloud.

Don’t Expect Too Much

As enticing as it may sound to generate dozens of “new” designs on the fly based on input text, there are considerable limitations of generative art tools. The end result is not always what a user has in mind and is typically difficult to customize. Each software has made its own stylistic assumptions which derive from the body of data available for training each model. Many AI art tools have trouble correctly spelling the exact input text or accurately depicting human hands. The current reality is that these tools cannot directly replace the talent and expertise of trained scientific and medical illustrators.

AI Artists to Follow

Countless artists have been creating generative AI art for years before the release of widely available software to the public. Artworks made with AI have reached one of the highest levels of exhibition in the art world, New York’s Museum of Modern Art.


In addition to the renown artists listed above, AIArtists.org hosts the world’s largest community of artists exploring AI.

Discourse on AI-Generated Art

© 2023 Anonymized Authors per NeurIPS Workshop Submission Policies [MIT License]