ROBOARTISTS FOR WRITING AND PAINTING:
THE ARTIFICIAL INTELLIGENCE THAT
COPIES THE GREAT ARTISTS

From year to year, artificial intelligence makes great strides and it seems that almost every day are presented new softwares and applications that exploit AI for ever different purposes and fields. Just think of the so-called machine learning, used in voice assistants, in automatic driving of vehicles, in chatbots and trivially in the keyboard of our smartphone that can predict what we are about to write based on what we have already written previously.

Automated learning uses a model of artificial “neurons”, inspired by the biological neural network, which seems to be able to find application even in disciplines previously considered exclusively a human competence such as writing and painting.

Creative Emulation: How Does GPT-3 Work?

During the “creative” process, GPT-3 exploits almost all existing content on the Internet . So, every time he gives textual output it is as if he were providing some sort of hypothesis based on some statistical calculations and the text available online made by the man he was trained with. This allows him, for example, to produce poetic texts respecting the style and vocabulary of famous authors . The OpenAI team tested this possibility by “asking” the algorithm to write a text as if it were William Shakespeare: the result is a text with a surprising and absolutely credible structure and vocabulary.

Same result for the test that allowed the Argentine scientist Manuel Aràoz to interview Albert Einstein without any problem.

The interview with Albert Einstein via GPT-3

The following is the transcription of an extract of the conversation between the scientist Manuel Aràoz and Albert Einstein, which took place through the GPT-3 algorithm .

Manuel: Albert Einstein. Pleased to meet you. I am Manuel Araoz and I come from the future to meet you.
Albert: Oh, really. What’s the future like?
Manuel: It’s incredible, actually. What year are we in?
Albert: We are in 1947. It’s January. This is Princeton, NJ. You are in America.
Manuel: Do you mind if I ask you a few questions about a book I’m writing? It’s about the great minds of history.
Albert: Go ahead. Ask me anything.

Manuel: What are you doing here in Princeton, in the United States. You were born in Germany, right?
Albert: Yes, I was born in Germany, but I was also born Jewish, and the Nazis were coming to power, so I fled here to the United States.
Manuel: What is your main occupation right now?
Albert: My main occupation is being a physics professor at Princeton University.
Manuel: So. Tell me more about how you came to a new gravitational model if you wish.
Albert: Okay. I wanted to explain the results of the Michelson-Morley experiment by building a new gravitational model, which would explain all the results of this experiment, and the fact that light always travels at the same speed in all inertial frames.

Manuel: Are space and time essentially the same thing?
Albert: Yes. In my theory, I have shown that space and time are related to each other in a very fundamental way. Basically, I have shown that how we perceive space and time depends on our relative motion.
Manuel: What would you say to people living in 2021, in about 75 years?
Albert: I would say that the puzzle of our existence has been reconstructed, but some pieces are still missing. We have a lot to learn. We should continue to explore and try to understand what those missing pieces are. What is most important is that we remain curious, for as long as possible.

Click here to read the full interview.

Rembrandt’s return between art and technology

Just as the GPT-3 manages to perfectly emulate the writing style of Shakespeare or Jane Austen, an algorithm born from the collaboration between Microsoft , ING Bank , Delft University of Technology , the advertising agency J.Walter Thompson and the Rembrandt House Museum has managed to make paintings as if it were a “new Rembrandt”. Blurring the boundaries between art and technology, The Next Rembrandt has collected and studied over 300 works by the Dutch artist trying to understand their style and distinctive characteristics and then create a whole new work, 3D printed using 13 layers of ink .

After classifying over 400 faces, the algorithm concluded that the subject would be between 30 and 40 years old, male and wearing black clothes and a hat. The next step was to develop software capable of “understanding” Rembrandt, based on his use of geometry, composition and pictorial materials. Microsoft then analyzed specific features such as eyes, nose, mouth and ears, mapping around 67 landmarks per face. The bot then learned how to create a “Rembrandt face” based on all this data, including the three-dimensionality of the brush strokes. Result? The work was so credible that it won more than 60 advertising awards .

Artificial intelligence, self-awareness and criticality

Painting works like these or the poetic texts produced by algorithms have aroused not a few criticisms in the art world and among its exponents: can paintings and literary works created by AI move as much as those created by human beings? Can programmers be defined as artists? Certainly, according to the current state of technology, algorithms and robots do not possess self-awareness or innate creativity, but “merely” act and react in response to the inputs and data they are subjected to during the learning phase. On this blog we have already wondered if robots could even become our best friends, but now the question that arises spontaneously is: is there still hope of preserving the creativity and inspiration of the human being or in the future even the creative professions can be “processed” with a code?