"Art is never finished only abandoned" -Leonard da Vinci
Introduction
The technology to create images using AI has exploded in both access and in popularity over the last few years. Now anyone with an internet connection can create images in a matter of seconds, anything they can imagine, in any style they want. Some people (probably not Leonardo da Vinci, I imagine) would consider these creations works of art. This blog will give you the basics of how they operate, briefly look at the history, and will explore the implications of such tech.
How it works
Algorithms are created to use databases of photos to compare existing photos and use their text descriptions to create new ones. A pair of neural networks go to work, one generates an image while the other grades and gives it feedback. The back and forth continues until the second network is satisfied. Each time the program is run, it gets a little better at its job.
These algorithms go to work when a user types in a description of their desired image. The technology scrubs the web looking for images with descriptions that match each word in the prompt. It then fetches, combines, and returns new images to the network for grading and feedback. It continues this process until the network technology deems it satisfactory and it is returned to the user. In moments, an image of anything you can imagine in anything style you want can be created.
A Brief History
While it may seem that this tech has come out of nowhere, it has a history dating back at least 50 years. Harold Cohen, who created one of the first programs called AARON in the 1960s, wanted the ability to create art through code. The technology evolved further with the development of generative adversarial networks (GANs) in 2014, which utilized the 'generator' and 'discriminator' pair of networks to develop its algorithm. Google's Deep Dream was another advancement in 2015 prompting the creation of apps that allowed users to transform their pictures into images that resembled works of art. Now some of the biggest players in the AI image game are DALL-E, Midjourney, and Stable Diffusion, just to mention a few. A quick web search will produce scores of platforms that create AI images.
Jason Allen’s A.I.-generated work, “Théâtre D’opéra Spatial,” took first place in the digital category at the Colorado State Fair.Credit...via Jason Allen source: https://www.nytimes.com/2022/09/02/technology/ai-artificial-intelligence-artists.html
Implications
Just like any other type of AI technology, it does come with its downsides. Other than the squandered hours that you spend going down a rabbit hole trying to figure out what The Dude would look like painted as the Mona Lisa when you should be doing other things (such as writing a blog), there are far more serious implications that can come into play. Many artists who spend years toiling to hone their craft take offense to people calling these images art. Also, having their intellectual property showing up as part of one of these images must be infuriating. There are many other implications as well that would be worthy of a deep dive.
Conclusion
Hopefully, the developers of this technology will use their powers for good and will be able to anticipate the negative effects that are possible and strive to use AI responsibly. This brief introduction is just the starting point of exploration of this topic and will be expanded on at some point. In conclusion, I can abide by Leonardo's belief, this art image needs a bit more work. The Dude's smile just isn't quite right.
Top comments (0)