They say “beauty is in the eyes of the viewer”, but does AI look at his art with the eyes?
I wonder if we can really measure how the artistic thing is using the matrix such as accuracy, precision, or remember? It may work to evaluate an image rating, but… art? In this place, things get vague. Nevertheless, such as the AI system such as Dell · E, Madjurni, and Stable Dispersion begins timelines flooding with real pictures and shining dreams, we want to know ourselves: Is it creative? And if it is (or who) get credit?
For example, take the famous portrait now “Edmund de Belmi”Who made headlines when it was sold in Christie for more than 432,000. The world’s first AI was marketed as an infield portrait, it looked like a glitch from another universe but was not in the real turn, it was in credit. The French collective behind the artwork was clear, which manufactured and sold the piece. But the original AI model? It was based on Ruby Barat’s open source code, a teenager AI artist whose role was not widely featured. In addition, the basic technology GANS was created by completely different researchers. So who really made art? An interesting study found the question and found out how the AI was described as a thinking agent or just a tool gives the form of how much credit people give it. When talking about AI as an artist, we give him more reputation, even if humans have worked mostly. It is a wild like giving a trophy to your paint brush.
So Even how do we start deciding the creativity of AI infiltrate art? This is still an open question. But I’m coming back Margaret A. Boden’s Creative Ability Framework that focuses on three important traits: Novelty, surprise and value. AI Creative Capacity has been classified:
- Common: Reying well -known ideas in new ways.
- Written by: Find out new variations in the current style.
- Change: Mold completely broken.
To measure them, researchers are using tools like:
- Lovelace 2.0 Test To see if the AI can produce compulsion but the results of the novel.
- Ritchie’s quality, face and spectacle Who examine the created results, appealing, appeal and creative value.
- And like the matrix Bisian surprise, spiritual distance, and styling deviation To determine how far an image is from known samples.
Personally, I think we will need even more suitable ways to evaluate AI infiltrate art. Something that is calculated Context, human joinAnd maybe even How much emotionally forced One piece is something that can imitate the current AI, but really doesn’t feel. I would not be surprised if we eventually look at the technologies such as the emotion or the biometric feedback tools that measure the human reaction by observing the art. Such data can present a new dimension to assess how effective or “successful” an artwork is really.
It also opens the door to new job titles: AI artists who are skilled in quick engineering, AI tool creators who develop models with embedded artistic logic or cultural antiques, and even AI art judges who make the machine -made creativity with the machine -made creativity.
Boden’s definition is still true to me: “Creative abilities have the ability to come up with ideas or samples that are new, amazing and valuable.”
And yet, I wonder if AI may not need to measure its art. The thing that attracts me the most is that it is born without seeing, yet somehow, we still see something in it.
To read more:
1. “Creative Ability and Machine Learning: A Survey”
2. “Understanding and creating art with AI: Review and Outlook”