I tried a hand at AI art recently, out of morbid curiosity and convenience for working around stock photo restrictions on a website. I also had seen some nice AI art online so I wanted to take a whack at it. The first subjects I fed the AI image generator were passable. It returned nice, generic images of budding cherry blossoms (see above). If you look closely it is a little plastic but lovely in its own right. Despite that, I didn’t end up using this image and opted for restriction-free classic art instead. Then I fed the generator a prompt of a shadow or outline of a mother and baby, hoping to conceal that it was AI generated this way. The results were okay, but I eventually discarded them and ended up using Mary Cassatt's art. The AI art was good enough to show the idea of a mom and baby, but I couldn't trust it to sell the message better than some original art could. For a discerning eye, the result didn’t have the authenticity we usually like family images to project. In fact, authenticity is kind of the point of selling a positive mother and baby narrative. If an AI image generator misses that point– well, then it has kind of missed the needs of an audience entirely. It also took a little bit of brain work to communicate to the computer what I needed. And there got to be a point where the thought that it would be easier, in a way, to make an image myself and have all the specifications exactly as I wanted and needed. Which, ironically enough, is not what AI art can effectively do. My takeaway is that using AI alone – at least for creative and communication work – is a little bit like asking a less qualified candidate to do a job for you. It can get the job done, but in the same way as if you hired a less professional and skilled person to do it. This is exactly what recruiters are sifting out when they normally hire people. In the long run, a bunch of unskilled persons will also likely lower returns for a company. In business, one of the most condemning things about AI is simply that people don’t like shelling out money for mediocre products or work. It takes a bit to make you buy something unless it is novel or useful. People will at times buy an item when it is well-marketed and has a good narrative around it – you see a lot of mass-produced art that has a generic style, even without it being AI art. But, it doesn’t really capture the heart of buyers in the same way a product with a novel human narrative has. At the root of all this buying and selling, humans subconsciously want to connect through creation to God. I think that AI almost seems to remind us of the disconnect. I don't expect good company leaders to entrust values and goals to AI. Best best-case scenario, it will help people do a quicker job of fixing mistakes on print images. But, maybe I am wrong and this generated cherry blossom image at the top of this post is really great and we can hand the goal of creativity off to a machine. I mean, the image is nice, but if you saw this everywhere wouldn't you eventually want to see a different type of art? Maybe I am not smart enough or trained enough yet to use AI to its full application. But, if that represents the average person’s dilemma then I think that, in the end, we are going to be okay. In the meantime, I am going to use Open Domain art from museums to convey a humanitarian message. (What is the point of this image being framed as such and with this backdrop? Conveying these questions to a machine is almost too much of a philosophical dilemma)
0 Comments
Leave a Reply. |
Quis ut Deus?In search of the Face of God. Personal blog with musings, thoughts, and stories. Archives
June 2024
Categories |