The Creativity Code

The Creativity Code: 

How AI is Learning to write, paint and think

Marcus Du Sautoy

Moving deeper to understand the capabilities and current state of the art in AI I made the choice to read the 2019 published book of Marcus do Savoy who is a professor of Math at University of Oxford.

I am astonished because of two reasons: 

1. It was published in 2019, meaning that all my impressive discoveries of the cutting edge innovation actually is related to almost 5 years ago; 

2. The beautiful and interesting narrative of Du Savoy shows real case scenarios and experiments in which AI is thriving to surpass human abilities to paint, think, and make art.

My favorite (and most disturbing) quote of this book was:

(...)”the reason why IP laws exist is to prevent others from using it and enabling the owner to generate a benefit. (...) What if AI creates a piece of art in the style of a living artist? It's likely that the programmer might be sued for copyright infringement but it is a very gray area. Inspiration and imitation are central to the artistic process. Where is the line between your own creation and copyright someone else? (...) Are coders the new artists?” 

The author makes reference to the literature of creativity in human beings which can involves the idea of creating creativity where he points out three different types of creativity:

1. Exploratory creativity in which the subject takes what is out there in a field of knowledge and explores new edges; 

2. Combination of two different things which didn't belong to the same field of knowledge, using rules of one field to reframe another field of work; 

3. Transformational creativity which is considered as a mysterious moments of game change establishing a new field of knowledge or rules (Picasso with cubism or James Joyce with Modernism are examples).

Is AI already there? Or more precisely was AI there in 2019? According to the book probably yes and the author gives some empirical examples on to provide grounds to his view: 

• Creation of a New Language: Google Brain pushed three algorithms to create a new sort of language using encryption methods, one of the algorithm should guess and the other two communicate between them, after some rounds actually neural networks were able to create a completely new way to communicate between them without human beings (and their algorithms) being able to understand what was being said. 

• Music and Literature Composition: Several examples are given to show the high quality surpassing human productions created by AI (that in 2019)! Most impressives were Cybernetic Poet, created by the futurist Ray Kurzweil and the notable project NaNoGenMo (National Novel Generation Month), which he initiated in 2013 as a playful response to NaNoWriMo (National Novel Writing Month). Instead of writing a novel manually, participants create algorithms to generate novels with at least 50,000 words, pushing the boundaries of authorship and creativity.

Those and other insights are provided by this beautifully narrative artistically written by a mathematician. 

Definitely worth reading!

João Pedro Paro - October 2024