Philosophers, neuroscientists, and artists have wondered a very long time about the nature of art. Attempting to answer the question leads us to many other questions, which may also remain totally unanswered. In many cases creativity is perceived as a property of the living meanwhile art is perceived as a property of the human race. At the scale of the living, even tiny creatures such as nematodes, made up of only 300 neurons, doesn’t only absorb food. A nematode also expresses a certain curiosity to explore its environment. If 300 neurons are enough to express curiosity, then it is not surprising to perceive the emergence of creative behavior in animal species made up of millions of neurons. However, although recent research shows the creative capacity of animals, we grant ourselves, as humans being, the privilege of artistic creation. We often think that we are the only species able to give birth to Art while experiencing aesthetic satisfaction from it. However, the artistic creation of many examples shows us that we are not the only species to produce what we call art.
Trending AI Articles:
1. Deep Learning Book Notes, Chapter 1
2. Deep Learning Book Notes, Chapter 2
3. Machines Demonstrate Self-Awareness
4. Visual Music & Machine Learning Workshop for Kids
Some bird species set up structures composed of small sticks resting on a round structure, that can be decorated with fruits, flowers and butterfly wings. Jared Diamond, one of the first researchers studying this type of structure, discovered that the ability of these birds to build such structures was in fact not completely innate. From an early age, they learned how to create the best structures by observing more experienced birds, but they also learned by their own experience and by making mistakes. If ever one of the elements of the structure is moved, they will replace the element in the exact place where it was previously.
Estheticians generally make a difference between the art generating an aesthetic appreciation from the works realized by the humans of that caused by nature. However, with the rise of artificial intelligence, we must now consider new forms of creativity, creation and potentially new forms of artistic creation. Those created by the machine.
“If we attach more importance to the emotion triggered to the audience, then we must ask ourselves if Human artwork is really different from the artwork resulting from artificial creativity.”
Let’s have a look at this work for example: depending on the angle of observation, it can create very different feelings, emotions or sensations according to the people. But as you have surely guessed, this artwork is not the result of the human imagination, but the result of Artificial Intelligence. The model able to generate these works was created by the AICAN research team. On their website, they publish many works created by their AI. According to them, the model was driven to answer the following question: If one teaches a machine the foundation of art and style and pushes it to generate new images that do not follow the previously established style, what will it be able to generate? The question we must ask ourselves is: where does the real value of art come from? Does it depend on the values and emotions that the creator wished to convey through his work? Or does it depend on the emotion that the work itself aroused to its viewer? If we attach more importance to the emotion triggered to the audience, then we must ask ourselves if Human artwork is really different from the artwork resulting from artificial creativity.
“If it would be possible to open the mind of someone at a particular location, the stored memory would be even more unique than a fingerprint”
What I always found very interesting in Humans is the differences in experiences while perceiving the same world. Two people watching the same movie will have had very different experiences and will build different emotions at each scene and recall memories of a unique past. If it would be possible to open the mind of someone at a particular location, the stored memory would be even more unique than a fingerprint. This chaotic evolution that every person is able to experience throughout his life influences his internal state at every moment of his life. So when we create something, with or without the intention of communicating information, our creation reflects at some point the state in which we are at time T, and the state in which we are depends on the present conditions but is also conditioned by all our past experiences.
On this point, it could be interesting to have a look at how the AI works. Most AIs, supposed to be creative, are connected by the principle of artificial neural networks. By taking inspiration from the way the human brain works, we create what we called artificial neural networks where each neuron is the result of an elegant combination of engineers, mathematicians and neuroscientists decisions of the behavior of each neuron, as well as the architecture of the entire network. Once the behavior is defined, the final model is more like a series of mathematical operations that allow the network to be trained to perform a given task. Just as one can train to perform a certain task, a network of artificial neurons can be trained to perform a specific task, such as recognizing what is on an image. What is interesting is that we can define other structures of artificial neurons that allow changing their state according to the current input, but also according to past experiences. This is called recurrent neural network. Thus, like our brain, the state of a recurrent neural network can be conditioned on the present information but also to all past information seen before. This architecture can be used for example to generate poems. Even if the results are nowhere near perfect, good progress is made every day until at some point, the generated poems will be indistinguishable from human poets.
We can also push even further the comparison between an artificial network and a biological brain. Some studies suggest that we are all able to consistently generate “divergent thinking”. In other words, it means that we are always thinking of many good or bad ideas. Finally, we would be able to generate from these ideas a thought that would stand out because it would make more sense than the others. This principle is called “Convergent Thinking”, it is done by what scientists call the EN or executive control network. the EN would thus be used to select an interesting idea from all the ideas generated. Of course, this would be done without being really aware of it, but finally, this moment of creativity where we begin to think or imagine something that seems interesting goes through a process of generation and selection. Here too, we can find similarities between the way the human brain works and the way we can generate the artwork that I showed you previously.
The artificial neural networks used for the creation of artistic works are called GAN, Generative Adversarial Network. These networks operate in two parts, the first one if the generator responsible for generating proposals, images in the case of the work shown previously. The second part is the discriminator responsible for the validation of the generator proposal. Over time, the discriminator becomes stronger in the validation of the proposals made by the generator, and for its part, the generator becomes better at generating proposals that make sense in order to be validated by the discriminator. Once the two networks are trained, we end up with the possibility of generating artwork. The generator is quite similar ultimately to the process of divergent thinking we were talking about, and the discriminator seems to be a great parallel to the present EN in the brain, used to extract good ideas from the bad ones. Of course, It’s not exactly the same process, the actual operation is actually not the same, but it does not preclude artificial neural networks from having their own way of generating content, and this content can prove to be totally relevant and produce good results.
When Einstein began to implement the principle of general relativity, his creativity allowed him not to generate art but mathematical formulas. Thanks to his highly enriched internal state, the mathematical formulas he could think of were enormously diverse, as was his ability to validate a reliable mathematical intuition. In the same way, the creative ability of artificial intelligence to generate content does not stop at the creation of artistic work but can potentially go much further. In the case of AlphaGo, the AI able to beat Lee Sedol, the Go game world champion, is not only renowned for its ability to understand the game but also for its power to develop strategies that no human had thought of it before. Although thousands of humans have already played. Yet when the sum of human cognitive abilities is gathered around the same point, we are statistically more inclined to find solutions and solve problems. It is thanks to what is called cognitive diversity. In the case of the game of Go, the sum of humans people had not been enough to implement a certain strategy implemented by AlphaGo.
“ What is perceived is not only artwork but a creative ability that can be used to help Humanity to solve some of his biggest issues.”
So concerning the artificial intelligence, the question is not so much whether the creativity of the AI can produce art but more if the AI can simply be creative to produce something we would not have thought of. This is the case in the strategy that the AI has developed in the game of GO. Adding the creativity of artificial intelligence to the whole of the human creative capacities can help us to find new solutions to the problems that we wish to solve. Today, and because we are only at the beginning, the creative ability of AI is much known to generate artwork, music, poems or to solve a game with a brand new strategy. But this creative capacity is also being used to help us to solve much more complex problems, such as problems with global warming, finding medical solutions, or problems directly related to the organization of our work in large-scale society. This is the reason why when you see artwork done by an AI, what should be perceived is not only artwork but a creative ability that can be used to help Humanity to solve some of his biggest issues.