Artificial intelligence (AI) has become quite the buzzword in the last few years. Whether it’s ChatGPT, Google Bard, Midjourney, DALL-E, Adobe Firefly, or any of the other hundreds of AI-based platforms, there’s no denying that AI is here to stay. One aspect of AI that I have been particularly fascinated by is AI image generators. While some of the resulting artwork is stunning, there is one key aspect missing from them all: the human element.
How Do They Work?
AI image generators are trained using machine learning. Basically, they scour an enormous library of content and continue to refine what they know about particular subjects and keywords. They also look for connections and patterns to further enhance their learning ability.
They’re like a masterful painter with a twist—a two-person team where one artist creates mesmerizing artworks and the other, an art critic, provides constant feedback for improvement. The painter (generator) starts by creating random pieces, and the critic (discriminator) carefully examines each one, guiding the painter to refine their technique. This artistic dance continues, with the painter learning from the critic’s observations and gradually produces breathtaking masterpieces that blur the line between reality and imagination. Together, they orchestrate a symphony of creativity, unveiling a gallery of awe-inspiring images that leave spectators spellbound.
To give you an idea of what AI can generate, that last paragraph was entirely written by ChatGPT. It’s becoming harder and harder to tell the difference between what is created by humans and what is created by AI. In fact, in 2022, a man used a piece of AI-generated art to win first prize at the Colorado State Fair. In 2023, another artist entered an AI-generated image into a major photography contest and won.
Think you can spot the difference between a photo and an AI-generated image? Take this short quiz!
What Does This Mean?
We are barely scratching the surface of AI capabilities. The fact that AI generators continue to refine themselves means they are just going to keep getting better. In just the past few years, the quality of what AI can create has improved astronomically.
However, since AI is evolving so quickly, there are still a lot of unknowns. Some of the main questions include:
How do copyright laws impact AI use?
- AI models learn from artwork that already exists, but since they generate something new, does that count as copyright infringement? This remains to be answered. One interesting tactic that Adobe has taken is limiting the database that its AI, Adobe Firefly, is trained on to Adobe Stock on Behance, a social media platform owned by Adobe. Since Adobe has the rights to all those images, you can be assured that no copyright infringement will take place. However, the downside is that the resulting images aren’t nearly as good as other platforms due to its limited data set, and the images look fairly sterile to me.
Will AI replace designers?
- There’s fear that AI will replace humans because of its ability to quickly create high-quality artwork at a fraction of the cost and time. A lot of this remains to be figured out, but I believe there will still be a need for human designers. AI models aren’t just sitting out there pondering the human condition and creating art based on it. They don’t get inspired or have life events that change their perspective on the world. Therefore, I don’t think that AI will necessarily replace designers, but those who are willing to utilize AI probably will.
At Moxie, we are exploring all kinds of applications for AI: inspiration, storyboarding, placeholder images on websites, mood boards, and more. We are excited about the possibilities AI provides and how it can improve our work and help us work more efficiently.
If you want to learn more about AI and how it pertains to images and artwork, I recommend the following video, “AI is Evolving Faster Than You Think Pt. 2,” (approximately 23 minutes long) from the YouTube channel ColdFusion.