Nuo Ma – Reflection 7


This is a very interesting reading topic, can computers generate art. It started with the early age of photography and portraitures. At their beginning, both goal is to honestly reflect what is seen, at this phase, their use of documentation is higher than their use as a form of art. Then with the revolution, there is argument if photography is art or not. Some say it is not because its machine rather than by human creativity. Second view is that it helps real artists as reference but should not be considered as equal to drawing and painting, third group think that it can eventually be a significant art form. History proved that photography is a major drive or catalyst for modern art.  Seen from today, photographs that only documents does not count for art, for those photographs that use special techniques, or reflect certain mood or event, or generally any picture that looks apart from ordinary pictures can be viewed as art. The author proposed that AI is like photography at the time, is an artistic tool of potentially creating art in the future. In the second part, the author discussed software art and its authorship, claiming that software executed on the artist’s instructions, the artist is the person who instigated and coordinated the work, not the software systems by themselves. However recent development in AI technologies has enabled some technology like GAN that can generate art further complicated this discussion. Should the author who created GAN the artists? Or those who write the code and running the optimization be? In my opinion, none of them. You see some GANs can learn certain painting styles and transfer onto any ordinary pictures, so that it looks like a piece of work from the original artist. Is this art? I think not, the painting styles originally developed by the painter is art. In my opinions, the value of the artworks can sometimes be judged by its scarcity and difficulty of reproducibility, although they are not major factors. Paintings like Mona Lisa are considered as valuable art, because there is only 1 painting, and no one would consider those being imitated by street side painters to be art, nor photo copied ones. The Sistine chapel is considered as valuable, because there is only 1 and impossible to reproduce. If AI easily learn the painting style and transfer onto another picture easily, it won’t be considered valuable, but rather a good tool to help artists. In the third part, the author proposed another view, that is art is an interaction between social agents, and social agents can create art, so until our current computers have personhood, they cannot create art. Another claim that I feel highly resonate about is that art requires intent or feeling to express something, such as emotion. Currently AI only identify some attributes of human artists, and then hypothesize that AIs with these attributes will be considered artists, but AI lack of intent, creativity and growth, all of which is required for a great artist. Then what about AI in the future? One cannot say for definite, but our current AI techniques and basic theories, for sure will not generate art or be considered as artists in the foreseeable time. However, should there be such an AI created in the future, the kind that public recognize as Artificial General Intelligence, then we can truely start to discuss if AI can create art.