Artists have always searched for inspiration in their muses. Actress and model Pattie Boyd was the inspiration behind songs written by as well George Harrison and Eric Clapton. Oscar Wilde’s devotion to Lord Alfred Douglas encouraged Wilde to create his most famous plays. William Shakespeare opened Henry V with a sincere appeal for assistance in creating a new work: “O for a Muse of fire, that would ascend the brightest heaven of invention.”

The fast-growing AI is a fantastic technology that can provide innovative ways of making however, it needs fixing and can mislead people and discredit the process of creativity. Image source: Dave Cutler (artist).

OPEN IN VIEWER Computers can be a muse. Can computers be an artist? Can algorithms make art?

Join PNAS’s alerts.

Receive alerts about new articles or receive an email when an article is mentioned.

MANAGE ALERTS For more information, check out the following article. Artists and authors are among the people trying out machine-learning computer models based on centuries of human art and producing their work at the touch of a button. From the highly-publicized chatbot ChatGPT to visual art generators such as DALL-E, research suggests AI (AI) can now mimic human creativity with a switch. It could be. When artists, researchers, and others evaluate the rapid advancement of AI technology’s capabilities and weaknesses, they’re witnessing an impressive technology that may provide creative possibilities. Still, it needs to be a better innovation that can mislead people and even undermine the creativity process. While specific experts point to apparent computer-generated creativity, others contend that AI technology can never be as effective as humans’ brains. “AI creativity is telling us more about our own creativity than anything else,” says Marcus du Sautoy, a mathematician at the University of Oxford (Oxford, UK) and the author of the book published in 2019. The Creativity Code: How AI Is Learning to Write, Paint and Think. “It is a new lens on the enormous creativity that we have created up to date.”Creativity is often a difficult concept to define. Can it be considered enough to allow a machine to change words if it doesn’t recognize the originality of these words? Does mimicking, modeling, and blending existing art styles constitute doing something completely new? When is a computer’s note-by-note demonstration of its mastery over the mathematical foundations of music go from boring to exciting? These are both philosophical and scientific inquiries. But, in the wake of rapid development and increasing curiosity about what algorithms can do, psychologists, mathematicians, and AI experts are attempting to answer these questions.

Flavors of Creativity

Creativity assessment, whether machine or human, requires some lateral thinking. Much like many experts in the field, Sautoy takes a concept seminal developed by Computer Scientist Margaret Boden in 1998, which breaks down creativity into three kinds ( 1). The first type, which Boden refers to as combinatorial creativity, is the fusion of ideas that are familiar to us. Generations of writers and poets have used this to achieve fame using a simple picture or an analogy. Imagine William Wordsworth and “I wandered lonely as a cloud.”

The second type of creativity, called exploratory, is a way to take what’s already there and expand the boundaries of what has been performed or observed. French painting master Claude Monet exploited new pigments to depict how light fell upon water lilies, and in doing so, he helped create Expressionism. The majority of human creativity is a process of exploration, Boden says. Her third branch of imagination is rare and more elusive. It is known as transformational creativity. It alters rules, alters the rules, and needs to be judged according to its criteria. This was a method of gaining temporary territorial control, as they anticipated that the game would shift to the middle in later moves. However, in the game’s 37th move, AlphaGo broke from conventionality and set its stone in the 5th line. It may seem like a lot. However, the commentators, as well as Sedol, were shocked. The game’s creators AlphaGo, realized it was performing something awe-inspiring when it calculated the probability of a human player performing this move as 10,000-1 against. It was a great move, and about 50 moves later, it tipped the scales and secured victory for the computer. “Transformational creativity is the tough one, where something new breaks the old system,” du Sautoy explains. “I would say that Move 37 had that quality because it challenged the previous playing system, with a radical new move.”

Under the Hood

How do the machines accomplish it? Even the most innovative algorithms can work only using the data on which they are taught. These machines, also known as generative AI, are available in a variety of kinds that use the lessons they have learned from training data in subtle, different ways.

Language bots such as ChatGPT generally employ a kind of neural network, also known as a Transformer, which discovers and then learns patterns of statistical significance in the sequence of words found in many pages of text online. To create an essay or poem, or slide show, It calculates the word that will be next to be, based on the words that came before (using all the millions of texts it has been taught).