Artificial Intelligence Art
Can artificial intelligence (AI) be an artist? Even if it is, can we call this art of artificial intelligence? Can artificial intelligence think like artists? What about feelings? We tried to analyze the relationship between artificial intelligence and art with all details and examples under the heading of Artificial Intelligence (AI) Art.
Artificial Intelligence (AI) Art
Artificial Intelligence Artists
Who knows, maybe one day artificial intelligence algorithms will create paintings, compose harmonious music, and perhaps write emotion-laden poems, imagining like an artist and combining what they’ve learned before.
But will the creation of artificial intelligence in the field of art make them a Claude Monet, a Mozart or a Edgar Allan Poe in the future? Can we call the art that artificial intelligence produces?
While the definition of art is controversial even today (which, according to some thinkers, is impossible to define art), it is very difficult to answer the question of whether artificial intelligence is the art they produce. But if technical perfection is sought in art, and perhaps a little bit original, it doesn’t seem that difficult for artificial intelligence to take our place.
Because in the face of artificial intelligence algorithms that minimize human-specific errors, it’s really hard for us to resist them with just our originality.
There is no guarantee that they will not gain this authenticity with the examples you will see in the continuation of the article.
However, when we consider the psychological effect of art, we can say that we can gain an advantage over artificial intelligence. More than the perfection of a work, we care that it has come out of the hands of a flawed person. What did the poet go through when he wrote them? Why did the painter not complete this work of art for many years? We see that these psychological effects are also a part of art.
From this point of view, I think that today’s artificial intelligence, which has no sense to convey, is decreasing in connection with art, and it increases the possibility that what they produce is not a work of art.
The Relationship Between Artificial Intelligence and Art
In today’s technology, we seem to be far from a powerful artificial intelligence capable of reasoning, imagining, emotional intelligence like humans.
Artificial intelligence is focused on achieving given/learned commands. And since there is no room for sentimentality, they complete the task without ever giving up. When it comes to such tasks, we can say that they have a marked success against humans.
For example: While professional lip readers can understand 20-60% of what is said, LipNet, an artificial intelligence developed by Oxford University, has a success rate of 93%.
It seems easy to measure the success of an artificial intelligence designed to do the job in the best way. Because, as in this example, we have data and we can say that the party that has a better percentage is the more successful party according to this data.
What if there are situations where being successful depends on the person? So what happens if art comes into play?
For example, who can draw the best human portrait? To you, Pablo Picasso. To me it’s Rembrandt Van Rijn.
So what if artificial intelligence algorithms are really successful in the field of art?
As a result, considering that there are artists who are influenced by each other and have produced works in the same movements, can’t artificial intelligence be influenced by some artists and produce their own works?
Examples of Artificial Intelligence in Music
Amper
An artificial intelligence-created and produced music album I AM AI was released in August 2017. Amper is an artificially intelligent music composer, producer and artist.
It was developed by a team of professional musicians and technologists and has the distinction of being the first artificial intelligence to create and produce an entire music album.
In doing so, Amper scanned through a large database of tens of thousands of songs, thus becoming capable of composing. You can listen to a piece of Amper’s lyrics written and composed here.
Magenta
Magenta is a Google open source research project that explores the role of machine learning in the art and music creation process.
It includes utilities to train machine learning models for the purpose of producing music with artificial intelligence algorithms, which include Python libraries and replace the source data with these libraries, and finally utilize these data to generate new content from these models.
They are working to produce new compositions and paintings with the deep learning and reinforcement learning algorithms they have developed. Although the compositions produced by the team, which are mostly known for their work on music, lack integrity for now, they are able to make music that sounds good when listened to in short pieces.
Here you can find a beautiful example produced by Magenta, as if it were created by a real DJ.
DeepBach
Johann Sebastian Bach, who lives in 18th century in Germany, is one of the most important representatives of baroque music with his unique beauty and technical competence in harmony.
DeepBach, one of the artificial intelligence studies in the field of music based on Bach’s works, has been developed and developed in Sony Computer Laboratories.
Although our knowledge of music is not enough, we need to look at the features of Bach’s works before talking about DeepBach artificial intelligence studies.
Because the DeepBach artificial intelligence algorithm has analyzed in detail more than 300 short polyphonic chants works by Johann Sebastian Bach. Bach’s works are based on Lutheran texts and consist of 1 melody–3 harmonies.
This simple but unique logical structure of Bach‘s works has attracted the attention of researchers for years and has been the subject of DeepBach‘s study of AI. The task of the algorithm was to create Bach-style harmonies when given a soprano melody.
The results of the study were tested on more than 1200 people, 25% of whom were professional musicians and music students.
Participants were played different harmonies and asked whether the harmony they were listening to belonged to Bach or an artificial intelligence.
DeepBach scored the highest in the study, which included harmonies produced by various artificial intelligence algorithms.
The reason why DeepBach is one of the most interesting examples of artificial intelligence researches in the field of art is the results of this study.
Half the participants thought DeepBach was a Bach. In another similar experiment, considering that Bach received 75% of the vote, including his own harmonies, it is indeed a very important achievement.
If you’re wondering what a Bach-style work by DeepBach is like, you can listen here.
Examples of Artificial Intelligence in Painting
Because painting is one of the most distinctive branches of art that is beautified by the touches of mathematics, artificial intelligence produces very successful results in this field of art.
For example, in the 15th century, the technique of perspective that artists such as Brunelleschi and Masaccio gained from painting using their knowledge of geometry enabled the rediscovery of the painting in a sense and provided the opportunity for many Renaissance artists such as Leonardo, Michelangelo, Rafael and Donatello to create unforgettable masterpieces.
AICAN
AICAN is an AI that has been developed in Rutgers University Art and Artificial Intelligence Laboratories and has achieved successful results in the field of painting.
AICAN is able to reveal its own innovative visuals by learning the existing styles and aesthetic elements in the paintings. We can say that the paintings that AICAN has created attract a lot of attention. People cannot even distinguish AICAN’s paintings from man-made works.
AICAN’s paintings are sold at auction worldwide. One of these paintings even found a buyer for $ 16,000.
There is also a virtual art gallery where you can review the artworks made by AICAN. It’s really amazing! I didn’t really know if I should be sad or happy that these studies came out of an artificial intelligence.
I would like to continue the artificial intelligence studies in painting with a very interesting example. GAN and CAN. In other words, Counterfeit and Detective artificial intelligences in the paintings.
GAN and CAN
In an article published in 2014, Ian Goodfellow of the University of Montreal and his colleagues introduced a kind of artificial neural network called GAN (Generative Adversarial Network).
GANs are basically composed of two neural networks that work for different purposes. The first one (generator) tries to produce fake data like a counterfeiter, and the other (discriminator) tries to compare whether the counterfeit data presented to itself by the counterfeiter is a fake.
In the example of painting art, since the producer’s neural network does not see the picture drawn by people before, it starts by drawing random lines and presents each picture it produces to the distinctive neural network.
The distinctive neural network (detective) can see real pictures drawn by people, comparing the drawings from the producer (counterfeiter) and the real pictures, listing the differences. The producer network (counterfeiter) also improves its drawings and re-submits them for approval, taking into account the feedback from the distinctive network (detective). In this way, the continuous repetitive flow ends when the discriminating neural network becomes unable to distinguish the drawings from the manufacturer.
Painting Made By Artificial Intelligence Sold For $ 432,500
How much do you think a picture made by artificial intelligence is worth?
If we consider it as an untouched output, it cannot be very valuable. But when AI is involved, I think it can change quite a bit.
Portrait of Edmond de Belamy was sold for 432,500 dollars at a sale in the Christie’s Auctions House, London’s most prestigious auction hall. (nearly 45 times its high estimate)
This portrait is the signature of the artificial intelligence program GAN, developed by Paris-based art collective Obvious. Indeed, in the bottom of the picture, AI has its own signature.
Portrait of Edmond de Belamy was the first painting made by artificial intelligence and sold in an auction house.
GAN has a database of 15.000 portraits made between 14-20 centuries.
DeepDream
DeepDream is a fan AI algorithm created by Google engineer Alexander Mordvintsev which uses a convolutional neural network to find and enhance patterns in images via algorithmic pareidolia, thus creating a dream-like appearance in the deliberately over-processed images.
Google’s deep artificial neural networks are trying to make the original picture more like what it thinks it is. Google calls this process Inceptionism.
Although it has been trained to distinguish more animal pictures for now, it is thought to be able to resemble different objects over time.
We tested this service (DeepDream Generator) offered by Google for free with Pierre Auguste Renoir’s Two Sisters (On the Terrace). The results are really interesting.
As you deepen the picture, you feel like you’re plunging deeper into a sweet dream or a nightmare.
Artificial Intelligence Examples in Literature
Since the works in the field of literature must have integrity from beginning to end, we can clearly say that literature is one of the areas of Art where artificial intelligence is the weakest at the moment.
For this reason, the scope of artificial intelligence studies in this field is generally kept short. But there are exceptions, too!
Artificial intelligence that can write novels!
A Japanese AI Almost Won a Literary Prize (The Day Computer Writes A Novel)
Hitoshi Matsubara, a professor of computer science from the Future University in Japan, announced that his research team’s short-form novel—co-created with an artificial intelligence—had passed the initial screening of a domestic literary competition.
And it wasn’t the only artificial intelligence novel to enter the competition.
The Hoshi Literary Prize received 1,450 novels, 11 of which were made with AI programs!
A novel written by artificial intelligence, The Day Computer Writes A Novel, reached the final. Although he couldn’t win the prize, its elimination of hundreds of novels and its rise to the final was of course a great success for the artificial intelligence world.
I wonder if we’ll get goose bumps when we read a poem written by AI. Or will they be able to write a gripping book that we will be so impressed with?
Or is the monopoly of emotional creativity solely on humans? Time will tell!
Benjamin (AI) Wrote a Movie Screenplay. (Sunspring)
Sunspring is the first short film written by an artificial intelligence! The world’s first automatic screenwriter, penned by Benjamin is a science fiction film.
How did Benjamin learn to write screenplay?
It has created its own screenplay, analyzing almost all science fiction scripts written to date. It examined the sentences and the words of the actors in these movies one by one and processed this detail to create a brand new short film. The only thing it failed when he wrote the screenplay was not to match the player names. Instead of using a name, it symbolized the characters of the film with letters.
It also analyzed over 30,000 songs and composed a song for use in the film.
What kind of movie is Sunspring?
When you watch the movie, you can see that different objects and events are sequenced. Suddenly, an artificial eye can emerge from the actor’s mouth, or while another player is dealing with a circuit board, a different love conversation can take place with another actress joining.
The actor finds himself among the stars with the song, and immediately answers an old cell phone that rings, all of which, as if taped to the wall, puts a very technological weapon in his mouth, perhaps thinking of suicide.
Immediately there is a black hole appears on the floor of the room, opening into the ceiling. Usually actors and actress say “I don’t know what are you talking about.”
You can click here to watch the Sunspring movie.
What is your opinion on Sunspring and future AI screenwriters?
Did you like the movie? Can AI be able to produce better movies in the future?
We would appreciate it if you would share with us your comments on what Sunspring and artificial intelligence produce in the fields of art.
As a result; it is truly incredible that an artificial intelligence can create such a film for the beginning. It is a proof that we can watch beautiful movies in the future.
Well, maybe one day, we can hear “and the oscar goes to…” sentences for an AI movie.
References:
- https://futurism.com/the-worlds-first-album-composed-and-produced-by-an-ai-has-been-unveiled/
- Bilim Teknik Dergisi December 2018
- https://www.youtube.com/watch?v=XUs6CznN8pw
- https://ai.google/research/teams/brain/magenta/
- https://www.youtube.com/watch?v=qFBQDfPyjoE
- https://magenta.tensorflow.org/
- https://www.youtube.com/watch?time_continue=17&v=QiBM7-5hA6o
- https://www.aican.io/art
- https://digitalspaces.io/demos/gallery/
- https://www.christies.com/
- https://medium.com/@devnag/generative-adversarial-networks-gans-in-50-lines-of-code-pytorch-e81b79659e3f
- https://deepdreamgenerator.com
- Barış Özcan – Yapay bir zeka sanatçı olabilir mi ? (Youtube Video)
- https://thenextweb.com