Is AI Art Real Art?

In this episode, we tackled a question that's been sparking debate and discussion: is AI-generated art “real" art? We explored the history of computer-generated art, the algorithms used to create it, and the ethical questions surrounding it.

Listen Now

A Look Back

It might seem like computer-generated art is a recent development, but it actually dates back to 1964, when a German mathematician named Georg Nees first used computers to generate art using basic shapes. In the early days, the art was produced using pre-computed formulas and algorithms, rather than through data-driven learning. A major shift happened around 2015 with the rise of deep learning AI. This allowed for AI to generate natural language captions for images and then, inversely, to generate images from text. Early attempts at generating images were low resolution, but technology has since improved rapidly, allowing for high-resolution, realistic images. OpenAI developed the first advanced AI art tool called DALL-E in 2021, but it wasn't released to the public until late 2022. Midjourney was actually the first to gain widespread adoption, releasing their model in July 2022.

The use of Midjourney in an art competition win sparked controversy, highlighting the discussion around AI art. Today, multiple companies have produced AI models that use diffusion techniques. The quality of AI-generated art is improving rapidly, making it increasingly difficult to distinguish from human-created art.

Demystifying Diffusion Models

The way these models work is quite complex and abstract, but they can essentially generate brand-new content. Diffusion models start with random noise and then, through a sequential process, transform that into meaningful images. They are trained on massive datasets of images and their corresponding text captions, learning to associate features of the image with words in the caption.

The diffusion process is like starting in a random spot in the middle of a desert with a compass and retracing steps to reach a specific location. Each time the path is different, resulting in a unique image. The compass in this analogy is the text prompt provided by the user.

These models have many features, which allow for a vast number of image combinations, making them capable of generating a diverse range of content. In addition to generating images from text, these models can also do things like image-to-image generation, inpainting and outpainting.

AI Tools for Art

Image generation tools include DALL-E, Midjourney, and Stable Diffusion.
Text-to-video models like Sora are also in development.
Music composition tools are also being developed by companies like Open AI, Stable Audio, and YouTube.

Ethical Concerns

AI models can reinforce harmful stereotypes due to biases present in the training data. These biases can be further exaggerated by how people are represented online. There is concern that current copyright laws are outdated for handling AI-generated art. Many artists worry that AI art tools could encourage plagiarism. A significant percentage of artists and the public believe that AI-generated art should not be considered art.

Is It “Real" Art?

AI art is often original and unique, as it's generated from random noise and is not directly copied from the training data.
The word “create" actually means “to form out of nothing" which aligns with the way these models generate images.
Some believe the absence of intention, imagination and emotion is what makes AI art different.
AI models are trained for speed and efficiency, which contrasts with the time and effort put into art by human artists.
AI art may lack authenticity and story.
There are arguments that the value of art lies in how much one is willing to risk to experience it.

Ownership and the Future:

The question of ownership is complicated, as it's unclear who owns AI-generated content - the user of the tool, the creator of the AI, or some other party. There are lawsuits in progress over copyright infringement regarding training data and mimicking artists' styles. Some models are starting to cite their outputs and there's a growing movement towards disclosing when content is AI-generated.

There are concerns about how AI-generated images may be used for misinformation. Many believe that AI art tools will be used to assist artists, similar to cameras, photo editing software and other tools that allow artists to make new work.

Final Thoughts

This episode has highlighted that the landscape of AI-generated art is complex and still evolving. As these tools improve and become more readily available, it is increasingly important to understand how they work, and what are the ethical implications. While there is not one simple answer to the question of whether AI art is “real" art, we hope we have provided you with some food for thought.

If you enjoyed reading, don’t forget to subscribe to our newsletter for more, share it with a friend or family member, and let us know your thoughts—whether it’s feedback, future topics, or guest ideas, we’d love to hear from you!