The Art of AI Image Generation: From Technology to Ethics
The blazing advancement of artificial intelligence (AI) has revolutionized various aspects of our lives, including image generation. With the advent of AI-powered image generation tools and services, the landscape of digital art and design has been transformed. In this blog post, we will scratch the surface of this technology and discuss some ethical concerns. Grab a cookie and let’s get started.
What the heck is AI Image Generation?
AI image generation is a rapidly evolving field of artificial intelligence that uses algorithms to create new images from scratch (or from other images). This technology has made it possible to create both photo realistic and stylized, visually stunning images. The potential applications of this technology are vast, including advertising, entertainment, fashion, and design. AI image generation is also becoming increasingly accessible to individuals, with various tools and services available online that allow users to generate custom images with ease.
Recent advancements in deep learning algorithms and neural networks have led to significant improvements in image quality and realism. By training these algorithms on vast databases of images, the AI can learn to recognize patterns and features in images and generate entirely new and unique visuals. This has led to an explosion of interest in AI image generation, with many artists, designers, and businesses exploring the potential of this technology to create new and exciting visual content.
These cookies were baked using AI technology.
Ok, cool, so how can I use this technology?
Riding on the wave of interest for this tech, many AI tools and services have popped up and gained popularity in recent years. Here are a few examples:
Midjourney can generate high-quality images in various styles, including abstract art, landscapes, and portraits. The tool uses advanced deep learning algorithms to create images that are both realistic and visually stunning. Users can access Midjourney by joining the Midjourney Discord server and typing commands to generate images. The process is still relatively straightforward, and users can customize various parameters to create unique and high-quality images.
Automatic1111 is a web UI that runs on your local machine and lets you input a ton of different settings to generate images using Stable Diffusion. Stable Diffusion is a latent diffusion model, a kind of deep generative neural network. Its code and model weights have been released publicly, and it can run on most consumer hardware equipped with a modest GPU with at least 8 GB VRAM.
Leonardo AI is a web service where you can generate images using textual prompts or other images. There are different pricing tiers that give you more generations and tools, but for now the free plan is enough to start experimenting. This is still in beta though, so in order to gain access you need to visit their Discord server and go through a couple of steps to get your account white-listed for use.
Great now I can create images of my dog being an astronaut
Yes… but hold on a minute, let’s first discuss some ethical ramifications of this technology.
One major concern is the potential for faking real people. AI-generated images can be used to create fake profiles, spread misinformation, or impersonate individuals. This raises questions about privacy, consent, and trust in the digital world. Moreover, AI-generated images have the potential to deceive people, leading to ethical dilemmas and the need for responsible use of this technology. Some concrete examples of these concerns are:
Donald Trump running from the police. Fake AI generated image.
Technology tends to evolve faster than laws and regulations can, and what we’re seeing right now is the beginning of a powerful technology being unleashed on society with little restraint. Some degree of chaos will ensue before society settles in the age of AI.
Training AI models on real artists’ work
Another ethical concern in AI image generation is the use of real artists' artwork as training data without their consent. Some models train on enormous datasets, millions, billions of images where proper consent from the original artists have never been given.
Some artists are upset that their work and style can be copied with ease, and they feel like it is a form of infringement on their intellectual property rights, especially when AI services earn money derived from their work. This led to widespread protests on art sites such as Artstation where users flooded the trending page (a common target for AI image “inspiration”) with “No AI” symbol.
One outspoken artist against the non-consensual use of their artwork for AI training is Sam Yang (@samdoesarts). While many people support Sam, there are those who think Sam should just accept reality and not complain. When you speak on an issue on the Internet, you also run the risk of becoming a target, something Sam had to experience when people started creating specialized models trained only on his artwork as retaliation to his cause.
Left: Original art by Sam Yang. Right: AI generated image from specialized model.
It's not surprising that many artists feel like their work is being stolen, but theft is perhaps not the correct word, do you remember the famous quote "You wouldn't download a car"? Is studying an artist copying her style or drawing inspiration from it, theft? Is taking a portfolio from an artist and feeding it into a machine learning algorithm theft? No, not theft, but certainly ethically problematic if the artist did not have a chance to consent.
Some argue that people have been copying and seeking inspiration from other artists since the dawn of human culture, and that AI is doing the same thing, just faster and on a bigger scale. While that might be true, the impact of this technology on society will be much more severe than that of the past. It’s not a black and white issue, there is a moral and ethical gray zone here that need to be explored further.
It's easy to draw parallels to other industries who had to battle technology, like the music and film industries when people started copying media and sharing it online. Eventually society adapts, and today we have great services for music and video media that also compensates the artists. I suspect the field of AI content will have to go on a similar journey.
While the advancement of AI image generation technology presents exciting opportunities, it also raises ethical concerns that need to be addressed. Responsible use of AI-generated images is essential, including obtaining proper consent, respecting intellectual property rights, and being mindful of the potential for misuse and deception. As AI continues to progress, it is important to have ongoing discussions and regulations to ensure that AI image generation (and other AI tools) is used in an ethical and responsible manner. This will take time, and I believe we're facing a period of wonderful, terrible, interesting, scary, joyful and hurtful chaos.
Mathias is a frontend specialist with a passion for creative artistic expressions in combination with technology. He enjoys both designing and building systems using the latest tools and frameworks. Outside of work he likes to dabble in digital art, do some gaming or eating cookies.