What’s the Ethical Trouble with Art Generators?

The landscape of art has witnessed a seismic shift with the emergence of artificial intelligence (AI). AI-powered art tools have revolutionized creativity, enabling artists and enthusiasts to explore new horizons. However, this technological marvel has also sparked a storm of debates, with AI art critics voicing concerns about its impact on creativity and AI ethics. As a blog in support of exploring this new landscape of AI and the tools it has to offer (a purpose which will remain) I’d be remiss not to address the elephant in the room.

Traditional artists fear that AI will devalue the artistic process and flood the market with mass-produced, algorithmically-generated works, diluting the significance of human-driven creativity. They argue that AI steals art, and that it lacks the soul and emotional depth that human artists infuse into their creations. While on the other hand, support for the “AI Revolution” involves combining the strengths of AI technologies and human capabilities rather than replacing one with the other. Whatever side you may fall on, AI ethics is a complex and evolving topic that demands thoughtful exploration and consideration.

Some artists and art enthusiasts have been calling for boycotts of AI-generated art (and AI in general), emphasizing the importance of supporting traditional artists and their unique contributions to the artistic landscape. They worry that if AI-generated art gains widespread acceptance and financial success, traditional artists may struggle to compete in an AI-dominated market. Their concerns are absolutely valid. As a traditional artist myself, I’m aligned with them.

Here’s the trouble: AI is here. It’s only getting bigger. Soon it will become as common to us as the internet itself. It’s touching so many corners of our world, for better or worse, and we have a long road ahead in understanding how to coexist with this technology. Resistance is essentially futile; instead, the key, like with anything new, is to learn, understand, and tackle the issues head-on.

Widespread adoption of AI technology is inevitable, and its potential benefits in various fields are too significant to ignore. In the realm of art and creativity, AI can assist and enhance the creative process for traditional artists, offering new tools and techniques that can complement their skills. Rather than shunning AI, embracing it as a tool for artistic exploration can lead to groundbreaking collaborations between humans and machines. That’s the ultimate idea behind this blog.

When digital art first emerged, I rejected it. I railed against it. It wasn’t “real art.” But it is. I know that now. People do amazing things drawing, painting and creating digitally. What I couldn’t accept back then was that digital art was a new tool, and when embraced, it enhanced the art world. AI is poised to do the same if handled correctly.

Understanding the ethical implications of AI in art is crucial. As AI-generated art becomes more prevalent, issues of intellectual property, attribution, and plagiarism will demand attention. Striking a balance between technological advancement and preserving the integrity of the artistic process will require ongoing dialogue, legal frameworks, and thoughtful regulations.

Education plays a pivotal role in navigating the complexities of AI’s impact on the creative world. Artists, consumers, and policymakers must stay informed about AI’s capabilities and limitations to make informed decisions. Ethical considerations, transparency in AI-generated art, and responsible use of technology are essential aspects to address. It calls for thoughtful introspection and open dialogue. Embracing AI as a creative tool while safeguarding artistic integrity is the path forward. Coexistence with AI demands an open-minded approach, focusing on learning, understanding, and addressing challenges together to shape a future where human creativity and AI innovation coexist harmoniously.

But for more clarity, let’s break it down.

The Creative Process: AI vs. Traditional Artists

1. AI-Generated Art:
   AI-generated art is created through machine learning algorithms that analyze vast datasets of existing artworks, photos, and other visual materials. These algorithms learn patterns and styles from the data and can produce new images based on these learned features. AI can mimic different art styles, from abstract to realistic, and create a diverse range of compositions. Proponents of AI-generated art argue that it represents a novel form of artistic expression, reflecting the collective input of the human-generated data.
2. Traditional Art:
   Traditional artists employ their skills, experiences, and emotions to create art through manual techniques such as painting, drawing, sculpting, or photography. Each traditional artwork is a direct product of the artist’s imagination and personal interpretation of the world around them. Traditional artists emphasize the value of the human touch, emotions, and the time-intensive process involved in creating each piece.

AI Ethics: The Art Conflict

One of the main critiques within the realm of AI art is the belief that AI art programs merely piece together existing images grossly violating copyrighted works without artists’ consent. The notion that AI art programs create collages of pre-existing images, is a misconception. (More on that later.) AI models like MidJourney, DALL-E, and Stable Diffusion learn to create in ways analogous to human cognition, constructing images rather than stitching them together. The AI’s creative process mirrors the essence of artistic construction, challenging the notion that it merely repackages existing content.

It is vital, however, to recognize that AI art’s ethical breach is rooted in the rapid pace of technological development. These AI models are trained on vast datasets, including artists’ works and copyrighted material, raising questions about consent and data usage. As technology advances, society must grapple with the ethical implications and ensure responsible usage.

The conflict between AI-generated art and traditional art revolves around questions of authenticity, ownership, and emotional connection. As AI technology advances, these debates are likely to intensify.

Concerns and misunderstandings around AI stealing copyrighted artwork stem from misconceptions about AI’s capabilities and its role in the creation and dissemination of art. While AI-generated art can indeed replicate certain styles and aesthetics, it cannot replace the originality and creative expression of human artists. Copyright infringement involving AI-generated content primarily arises from human actions and not from the AI itself.

So let’s proceed into those complex waters for a moment:

How AI Image-Generation Actually Works

AI machine learning for image generation typically involves using a type of deep learning model known as a Generative Adversarial Network (GAN). GANs consist of two neural networks, the Generator and the Discriminator, which work together to create and evaluate synthetic data, such as images.

Here’s a high-level overview of how AI machine learning works in GANs for image generation:

• 1. Generator Network: The Generator network’s primary task is to generate synthetic images from random noise. It starts with random input data (often referred to as latent vectors) and processes it through multiple layers of the neural network to produce an output image. Initially, the generated images are usually random and of poor quality.
• 2. Discriminator Network: The Discriminator network serves as the “critic” and tries to distinguish between real images (from a dataset) and fake images generated by the Generator. The Discriminator is also a neural network, but it is trained on real images and fake images produced by the Generator.
• 3. Training Process: During the training process, the Generator and Discriminator networks play a game, where they try to outsmart each other. The Generator aims to generate images that are so realistic that the Discriminator cannot distinguish them from real images. On the other hand, the Discriminator aims to become more accurate in distinguishing real from fake images.
• 4. Adversarial Learning: As the training progresses, both networks get better at their respective tasks. The Generator improves its ability to generate more realistic images, while the Discriminator improves its ability to differentiate between real and fake images. This adversarial learning process continues until the generated images become indistinguishable from real images to the Discriminator.
• 5. Convergence: When the training converges, the Generator becomes skilled at generating highly realistic images that resemble those from the training dataset. At this point, the GAN has learned to generate new images that share similar characteristics and patterns with the original dataset.
• 6. Variations: Different variations of GANs, such as Conditional GANs (cGANs) and StyleGANs, have been developed to introduce more control and creativity into the image generation process. cGANs, for instance, allow the Generator to generate images based on specific conditions, while StyleGANs enable the generation of diverse and high-resolution images.

In summary, AI machine learning in image generation using GANs involves training a Generator to create synthetic images while simultaneously training a Discriminator to distinguish between real and fake images. Through adversarial learning, the Generator becomes skilled at producing realistic images, resulting in AI-generated images that closely resemble those from the original dataset.

AI Ethics: Ethical vs Unethical

Both unethical AI and ethical AI can have implications for art-generation in terms of copyright infringement, but they differ significantly in how they approach and handle the issue.

Unethically Trained AI and Copyright Infringement:

Some AI image generators use existing copyrighted artwork as part of their training data. These generators often collect large datasets of images from various sources, which may include copyrighted artworks, photographs, and other visual content. They then use this data to train the neural networks responsible for generating new images.

The issue with using copyrighted artwork in this manner is that it raises legal and ethical concerns related to intellectual property rights. If the AI generator uses copyrighted artworks without proper authorization or licensing, it can potentially infringe upon the copyrights of the original creators. This unauthorized use of copyrighted materials can lead to legal repercussions for the developers of the AI system.

1. Unethical Data Usage: AI may be trained on copyrighted artwork without proper authorization, leading to the creation of derivative works that infringe upon the original artist’s copyright.
2. Plagiarism and Replication: AI may be programmed to replicate existing artwork without giving credit or proper attribution to the original artist, leading to plagiarism and copyright violations.
3. Commercial Exploitation: Some AI systems may produce art that is intended for commercial purposes, further exacerbating copyright infringement issues, as the generated works may be used without the original artist’s consent for commercial gain.
4. Non-Respect of License Terms: AI might ignore or bypass the licenses associated with original artwork, such as Creative Commons licenses, and generate art that does not comply with the terms set by the copyright holders.

Ethical AI and Copyright Compliance:

On the other hand, organizations like OpenAI and Adobe follow ethical practices in AI development. They typically use large datasets of images that are either in the public domain, have been released under open licenses, or have been explicitly authorized for use in AI research. These datasets are carefully curated to avoid any copyright or legal issues.

OpenAI’s approach involves machine learning, particularly deep learning models like Generative Adversarial Networks (GANs) or Variational Autoencoders (VAEs), to generate new images. These models are trained on the curated dataset, and their neural networks learn to identify patterns and features present in the images. Once trained, the models can generate new images based on the learned patterns, allowing for creative and original artwork generation.

The main difference is that OpenAI and similar organizations prioritize ethical considerations and adhere to copyright laws by using legally obtained datasets for training. As a result, their AI image generators produce new content that is not based on specific copyrighted artworks and, therefore, do not infringe on any intellectual property rights. That’s the claim anyway. As we speak, there are lawsuits flying around that we as a society will methodically plod through, our future with AI set to be an evolution on all fronts.

Still, the ideals for ethical AI are as follows:

1. Respect for Copyright: Ethical AI is designed to respect copyright laws and principles. It is programmed to avoid generating art that directly copies or infringes upon copyrighted works.
2. Creative Inspiration: Ethical AI can be used to assist artists in their creative process by generating art that is original and serves as a source of inspiration, rather than simply replicating existing copyrighted artwork.
3. Attribution and Fair Use: Ethical AI can be programmed to provide proper attribution for any external content used as part of its art-generation process. It can also consider fair use principles when generating art that might involve elements from copyrighted works.
4. Educating Users: Ethical AI platforms can educate users about copyright laws and the importance of respecting intellectual property rights, fostering a culture of responsible art creation and consumption.

In summary, the key distinction between AI image generators that have sampled copyrighted artwork and ethical organizations is in the sourcing of training data. Ethical AI developers ensure they use legally obtained datasets, avoiding any copyright issues and respecting intellectual property rights.

As the technology evolves, it becomes crucial for AI developers, artists, and users to adhere to ethical guidelines and legal frameworks to ensure that AI-generated art respects copyright and fosters a healthy creative ecosystem.

Why Creatives Need to Learn AI

Denial and ignorance are not going to help anyone in this situation. AI is not going away, and waving pitchforks is not going to make it. As AI continues to play an increasingly prominent role in the art world, it is crucial for artists to stay informed about AI’s capabilities and limitations. By understanding the nuances of AI-generated art and taking proactive measures to protect their original works, artists can confidently navigate the evolving landscape of AI while safeguarding their artistic rights. (You can actually check to see if your work has been used to train AI here.)

First and foremost, AI is a tool. In many cases, it’s a tool for creators. And like other innovative tools that came before it (digital art anyone?), AI faces great friction in the face of great change. Did you know that photography was once considered the pariah of the art world? The idea that someone could capture an image at the click of a button was a blasphemous threat to artists who slaved in traditional mediums, honing their craft. But as time shows, photography became its own art form, apart from others. And the existence of photography did not mean the extinction of art in other forms. It was something new.

So too is AI.

The nature of learning a craft–an art–is to learn it. Learn from the masters. That’s what the AI is doing at an incomprehensible level because it’s a machine, and that scares people. It fuels controversy. And is the price of innovation, for better or worse. (Just like that thing we all love and hate–the internet.)

As a traditional artist and writer, I would never advocate for AI to replace human creatives. I do not agree with those who are trying to replace creatives, thinking that AI will pave a way for complete control over a creative landscape. It doesn’t work that way. If there’s one thing I’ve truly learned in my AI journey so far, it’s that you still need the humans even with the AI. And you need them in a big way. Joke’s on you if you think you don’t. The AI has limits.

I do, however, advocate for human creatives to embrace AI as a tool in their creative repertoire. Don’t be scared of it. Learn it. Learn it especially if you feel threatened by it. Demystify it, understand it, and flip the narrative. Use it to your advantage so it can’t be used against you. Remember the people who didn’t want to learn about the computer when that came around? No, of course you don’t. And like the computer, AI is going to create as many jobs as it’s going to affect. So learn it.

Instead of shunning AI, embracing it while understanding its capabilities can empower artists to protect and preserve their art. Furthermore, collaborating with AI as a creative tool rather than fearing it as a competitor can open new possibilities for artists, enabling them to embrace technology while preserving the uniqueness of their artistic vision.

Closing Thoughts

Beyond the surface-level critiques, deeper existential and philosophical concerns emerge when we delve into the essence of AI art. The ability of AI programs to assimilate vast amounts of data, influences, and artistic traditions to create outputs touches a sensitive nerve within humanity. The notion that machines can perform what was once exclusively human—tapping into the realm of artistic expression—triggers profound existential questions.

AI art critics’ concerns about ethics may be entangled with an emotional upheaval, considering the awe-inspiring ability of machines to now engage in a process often regarded as uniquely human. This feeling of existential dread is understandable, even among the staunchest advocates of AI. However, it is essential to recognize that AI art’s capabilities are not intended to replace human creativity but rather to complement and augment it.

The contention that AI art tools draw from artists’ works without consent denies the nature of artistic endeavors and creative inspiration. Artists throughout history have drawn upon countless influences, images, and traditions to shape their craft. Whether stumbled upon in museums, books, universities, or online, these inspirations have always fueled creativity. We artists have always learned from those who have come before us, with or without permission. At its root, the machines are now doing the same.

Similarly, plagiarism existed long before AI entered the picture. The act of building AI models specifically on an artist’s work may constitute plagiarism, but this issue predates the technology’s advent. AI art tools are not the source of plagiarism; they merely reflect the intentions of their users. The onus of AI ethics then falls upon the user as well as the developer–the humans.

As AI art critics voice their concerns, the debate surrounding AI art is bound to continue. The fear-driven backlash against AI art and its advocates may hamper constructive discussions about AI ethics and responsible usage. The key lies in understanding that AI art tools are here to stay, and they hold the potential to revolutionize art for the better.

Boycotting AI is not the solution to protect traditional art; nor is it realistic. A boycott is not going to stop the progression of AI anymore than it would have the advent of the internet, or the computer, or any other innovation that had people rallying against it in the streets. It’s going to be a long road to coexistence with this new technology, replete with many questions of ethics, many legal cases, many debates and discussions–we’re only just getting started. It is essential, however, to understand AI’s potential and harness its power responsibly.

Rather than ostracizing AI art, artists and enthusiasts could benefit from welcoming and engaging with it. Embrace AI art as an ally that opens new vistas of creativity and possibilities. Transparency, consent, and thoughtful discussions must guide the ethical use of AI art tools, paving the way for a harmonious coexistence of human art and AI-generated art.

*The images in this article were generated in Midjourney.