Earlier last year, I talked to a former teacher of mine about my plans of making generative art with a pen plotter. She was not amused. I didn't understand, she elaborated. During that time, the writer guilds were striking in Hollywood, among other things against the use of ChatGPT for generating scripts. And making "generative" art sounded to her like I was invested in creating some sort of model that fed on other artists' work without compensation them and finally replacing them. I also failed to give her an understanding, that despite the term, I was doing nothing of the sort.
One of the reasons I failed was definitely that I was short of terminology. Also a quick google search gave me no easy and distinct answer, rather many opinions and contradictions. So I unironically asked ChatGPT as expert witness for distinction of the three terms computational, generative and algorithmic art, and how they relate to text-to-image models. I also asked it about the threshold of originality concerning all of these.
While the answer makes sense and feels correct, and probably is not completely off, I have a bitter after taste about the experience. A google search that does not reveal a definite answer means: it's not that simple. After all it takes time and discourse to shape terms. And this is portrayed in the search result. ChatGPT distills the statistically most likely answer out of this question. I noticed, how this answer feels more correct in my head, just because it was easier to read and more comprehensive then a quick google search.
ChatGPT provides a new form of reduction and simplification. While reductions of complexity are necessary, they are not easy to do and still often stay in people's memories as the full thing. Human made reductions usually come from a conscious compromise, ChatGPT is based on statistical models. I can't say yet if this is bad by itself, but it certainly can't hurt to keep that in mind. Especially the fact, that the comprehensiveness and eloquence of an answer does not make it the more correct answer.
Ironically, you can read the output as interview below.
What is the difference between Computational, Generative and Algorithmic Art?
Computational, generative, and algorithmic art are related but distinct fields within the realm of digital and computer-based art creation.
Computational Art: This refers to art that is created using computational processes or systems. It involves the use of computer programming, software, and technology as integral parts of the artistic creation process. Computational artists often write code or use algorithms to generate visual or interactive art forms. This field encompasses a wide range of artistic expressions, from algorithmically generated images to interactive installations driven by software.
Generative Art: Generative art is a subset of computational art that specifically focuses on creating art that has some degree of autonomy or is self-sustaining. Artists in this field write algorithms or create systems that produce artwork autonomously, often with a degree of randomness or variability built into the process. Generative art often explores the relationship between the artist, the code or system they create, and the resulting artwork, sometimes allowing for emergent or unpredictable outcomes.
Algorithmic Art: Algorithmic art emphasizes the role of algorithms in the creation of artwork. It involves the use of step-by-step procedures or rules to generate artistic output. This can include visual art, music, or other creative forms. Artists in this field design algorithms that dictate how elements are arranged, combined, or manipulated to produce the final artwork.
In essence, computational art is a broad term encompassing art created through computational means, while generative and algorithmic art are specific approaches within this domain, focusing on self-sustaining or autonomous artwork and the use of algorithms, respectively. These fields often overlap, and many artists may utilize elements from all three in their creative processes.
How does generating images with text-to-image models relate to the above concepts, especially regarding threshold of originality?
Generating images with text-to-image models like OpenAI's DALL-E or similar systems involves using machine learning algorithms to create visual content based on textual descriptions. These models are a fusion of natural language understanding and image generation capabilities, often leveraging large datasets to learn the relationships between words and visual concepts.
Regarding the concepts of computational, generative, and algorithmic art:
Computational Art: Image generation through text-to-image models falls within the realm of computational art. These models use complex computational processes, such as deep neural networks, to generate images based on input text. Artists or users provide textual descriptions, and the computational system interprets and generates corresponding visual representations.
Generative Art: Text-to-image models can be considered a form of generative art as they autonomously create images based on textual input. The models have the ability to generate diverse, sometimes surprising, visual outputs from descriptive text, showcasing an element of autonomy and creativity.
Algorithmic Art: While text-to-image models use complex algorithms and neural network architectures, they might not strictly adhere to traditional algorithmic art principles where specific step-by-step procedures or rules are designed by the artist. However, the underlying functioning of these models involves algorithmic processes to interpret language and generate visuals, making them relevant to the algorithmic art domain. Regarding the threshold of originality, it's an important legal concept in copyright law. In art created using text-to-image models, the question of originality can be complex. While the AI system generates the visuals, the input and guidance come from a human. Whether the resulting image meets the threshold of originality, and hence qualifies for copyright protection, may depend on various factors including the level of creativity involved in crafting the input description and the degree of human involvement in shaping the final output.
As of now, legal and philosophical discussions are ongoing to determine how copyright laws apply to AI-generated art. The involvement of human creativity in generating input, curating outputs, or providing direction to AI systems remains a critical aspect in determining the originality and copyright of AI-generated artworks.
How does the threshold of originality apply to works created with computational methods but without text-to-image models?
The threshold of originality is a crucial legal concept in copyright law that determines whether a work qualifies for copyright protection. It assesses the minimum level of creativity and originality required for a work to be eligible for copyright.
In the context of works created using computational methods (but without text-to-image models), the application of the threshold of originality remains pertinent. These works may include computer-generated art, algorithmically generated music, procedurally generated content in games, and more.
The determination of whether such works meet the threshold of originality can be complex and may involve several considerations:
Human Creativity and Input: Even though computational methods are used to generate these works, the degree of human creativity and input involved plays a crucial role. If the process involves substantial creative decisions or inputs from a human creator, it's more likely to meet the threshold. Level of Automation vs. Human Intervention: If the computational process is entirely autonomous, with minimal or no human input beyond setting initial parameters, it might be challenging to establish the required level of human creativity for copyright protection.
Artistic Expression and Uniqueness: The uniqueness, originality, and expressive elements of the resulting work are also evaluated. If the output demonstrates distinctive artistic expression, deviation from generic outputs, or reflects the creator's personality or style, it could support meeting the threshold of originality. Complexity and Novelty of the Computational Process: The complexity of the computational process used and its novelty can also influence the determination. Innovative algorithms or unique computational methods that produce unexpected or novel outcomes may contribute to meeting the threshold.
In summary, the application of the threshold of originality to works created through computational methods involves examining the extent of human creativity, the level of human input, the uniqueness of the output, and the complexity of the process. Courts and legal systems often evaluate these factors on a case-by-case basis when determining copyright eligibility for works created using computational means.
Do you think an image created with a text-to-image model can ever have the same threshold of originality as, let's say a painting by Picasso?
The concept of originality in art is nuanced and can be influenced by various factors, including creativity, human intent, and cultural context. Comparing an image created by a text-to-image model to a painting by a renowned artist like Picasso involves considering several aspects:
Creativity and Intent: Picasso's paintings are a product of his creative vision, artistic skills, and intentional expression. They reflect his unique style, emotions, and ideas. On the other hand, images generated by text-to-image models rely on computational algorithms trained on datasets and lack a direct emotional or intentional human expression. Human Involvement: Picasso's paintings are a result of his personal decisions, experiences, and artistic choices. In contrast, while humans provide textual prompts for text-to-image models, the actual image generation process is primarily driven by computational algorithms without direct human intervention in the creation of each pixel or stroke.
Cultural and Historical Context: Picasso's artworks often carry significant cultural and historical value. They are part of a larger narrative within the art world, influenced by the artist's time, movements, and societal context. AI-generated images, while innovative, lack this historical and cultural depth. Considering these factors, it's challenging for an image created by a text-to-image model to achieve the same level of originality as a masterpiece by Picasso or other renowned artists. While AI-generated art can be impressive and innovative, it lacks the deeply personal and intentional human expression, artistic context, and historical significance associated with works by acclaimed artists like Picasso.
However, the perception of originality and artistic value might evolve as technology progresses, and societal attitudes toward AI-generated art shift. Debates around the role of AI in creativity and its place in the art world continue, raising questions about the nature of authorship, creativity, and the definition of originality in the context of machine-generated art.
Thank you for your time.
You're welcome! If you have more questions or need further assistance in the future, feel free to reach out. Have a great day!
Support this work or buy my art works over at