Alexei Grinbaum, Ph.D., HDR, is a physicist and philosopher at LARSIM, the Philosophy of Science Group at CEA-Saclay near Paris. His main interest is in the foundations of quantum theory. He also writes on the ethical and social aspects of emerging technologies, including robotics and artificial intelligence. Grinbaum is a member of the French national ethics committee for digital technologies (CNPEN) as well as of the French ethics commission for research in information technology (Cerna). His books include "Mécanique des étreintes" (2014) and "Les robots et le mal" (2019).
Should we give different definitions to contemporary art produced by humans and by machines? Is the stress on contemporary important for you?
I don't think art, the concept of art applies to anyone except humans. Art is a part of the human condition for me. So whether that art is produced by machines, whether the art is produced by humans, whether that art is produced or rather was produced by humans who are dead, it's art from the point of view of those who are looking. Only a human spectator, only a human user will be able to call that art. Art doesn't grow on trees, art is not in space − art is in the human mind.
I've never been comfortable with the word contemporary. I don't think there is such a thing as contemporary art versus art that doesn't make sense to me. So there is art made by humans, including dead humans. Surely some art by dead humans wasn't even intended as art.
You can try to define art as a piece of work that is completely useless, that is made just for the point of being contemplated or existing. There is no utility in there.
Machines can produce behavior that will be read by humans as intentional behavior so they can imitate that, they can create a very convincing imitation that no one will be able to tell from intentional behaviors. Whether the machine imitates intentional behavior or not whenever it makes, it needs to be perceived by a human in order to have aesthetic value for the human.
You can ask yourself, what is the aesthetic value of this or that ancient sculpture or text. Today in 2021 this aesthetic value is very likely totally different from the original intent of the human 2 000 years ago.
Is art necessarily connected with the notion of creativity? If yes, please provide a definition of creativity. If not, then what is the “engine” of art (e. g. you consider art as a spiritual practice or a natural part of life and industrial production as Soviet productivists did etc.)?
I distinguish between the verbs to create and to produce. To make, to produce and to create is manifold. So I like the distinction from Hannah Arendt between acting willfully and making or producing. Definitely, art is an act of creation into the world, bringing something into the world and not of production; production or manufacturing are different things.
The very idea of bringing a meaning into the world, to me qualifies a creation, not a production. A spectator or a reader may have their own meanings, but a work of art brings meaning into the world that doesn't have to be objective, it can be fully subjective.
On the contrary, machines do not bring meanings into the world. They can write beautiful poems, but they have no idea about any meaning of that poem. Only when a human reads it and says, “Wow, how unexpectedly beautiful!” or “How surprisingly ugly!”. But that is done by humans.
Do you see any medium limitations in machine-produced art? In perceiving machine produced art?
You can create a whole world where there will be artificial agents producing things and other agents saying that certain things are beautiful or not beautiful (as humans actually can get the meaning of what they are doing, what's happened to them, from social practice). There will be certain problems there because very likely this aesthetics will not be recognizable by humans in any way. Why is that? Because we as humans make our judgment based on a number of motives, so our visual system works based on the analysis of symmetry and edges, machine vision works based on the analysis of texture. So if we distinguish a number of motives that are human like symmetry for visual perception or, machines have no idea of which motives, which fundamental motives are human and which motives are not human.
How do you see machine artists’ ethic code if both — humans and machines — could contribute to it? Do you imagine machine artists’ impact on human artists’ in a way like an authorship vs cc, high res vs circulationism etc.)?
It also applies to ethics. A number of machines can create a world with different kinds of judgments within that world. In that sense, it would be nice to create a bunch of adversarial networks of GANs that sort of develop their own aesthetics within that world. Certain GANs will be producing works of art and other GANs will be making judgments. Maybe humans won't even understand anything there.
I think nobody has really done this at such a scale that there can be a real small world of aesthetics, of machine aesthetics. You open it up and you suddenly see that machines have chosen as beautiful. Certain things will be similar to the human ideal of beauty and certain other things will not. I think imitation can go very, very far there.
I formulate ethical requirements in terms of responsibility, not in terms of legal issues or limits not to cross. You can say that the sculptor who made a sculpture of a wage slave trader was so bad. Is that sculptor somehow responsible for immortalizing awful in sculpture? I don't know. It is related to our changing values and perceptions as a civilization, and that will always be going on, changing again and again. I am not saying that this creates a responsibility for that sculptor who is dead. There is very much responsibility on us − the ones who keep these items, who look at them.
Digital ethics for artists is a limit or a bond on what can be done. I would not put a limit on art. You make a piece of art and there I am not your censor. You are the artist you are free to do what you want. Although if you release your piece of work, you show it to the others, it becomes part of our common world and that is when you become responsible. And it's not a question of legal responsibilities or liability, as a free artist rules the burden of that change that will happen to the world through his work. So I would very much place that responsibility on the artist, but not translate that responsibility into any limitation or constraint on what can not a priori.
A machine can create something that will be seen as beautiful one day, and the next day it will suddenly turn very ugly. The machine will not take any responsibility for that.
We have to design specific procedures in order to extract these NHAs from moral judgment because they are not moral agents and moral judgment on them can only be done by projection. Randomness, the use of chance is one of the procedures that extracts NHAs from the projection of morality that is not theirs.
The big change (that comes unnoticed) comes through mutual imitation, double mimesis, but not competition. People are learning from machines. We are not human in the same way we were human in 1850 or 1950.
Find more information about Alexei Grinbaum