The Creativity Code: How AI is learning to Write, Paint and Think [Part 1]
“For a machine to be deemed truly creative requires one extra step: its contribution should be more than an expression of the coder’s creativity or that of the person who built the data set. That is the challenge Ada Lovelace believed was insurmountable” (Du Sautoy, M., 2019).
Written by Marcus Du Sautoy, distinguished professor of mathematics at Oxford University, the title of the book is more than self-explanatory; to my mind, it is one of the most original pieces on Artificial Intelligence. Du Sautoy surprises the reader with an entirely novel approach. Not only does he focus on the most neglected and under-researched aspects of AI, Sautoy offers the reader a literary and historical perspective making the book one of the most immersive and accessible accounts of AI. We will be diving into this book for the next several days.
For years it has been accepted that true novelty is an empty dream when it comes to algorithmic processes. “You cannot get out more than you put in” — was the usual mantra. However, with modern bottom-up programming methods it seems that AI will be exhibiting some human-like traits after all. What does it mean for an AI to achieve a level of complexity that brings it close to a human being? For Sautoy, the secret lies with the notion of creativity.
“We have this extraordinary ability to imagine and innovate and to create works of art that elevate, expand and transform what it means to be human. These are the outpourings of what I call the human code” (Du Sautoy, M., 2019).
Identifying the human code means identifying the creativity code. This is what we are after. It is what accounts for human ingeniousness across the arts, sciences and the humanities. The creativity code is more than mere novelty, it refers to the ability to make something original and valuable. Something that could even surprise the author herself. But is there a code for that at all? We know for a fact that certain neuronal and chemical processes underlie this activity, but can these operations be isolated and reproduced under controlled conditions?
“Could our creativity be more algorithmic and rule-based than we might want to acknowledge?” (Du Sautoy, M., 2019).
To steer clear of unsubstantiated claims, like the ambition of modelling the next Mozart, Goethe or Alan Turing, Sautoy wants to investigate whether we could program an A.I. to exhibit basic levels of creativity. The type of inventiveness that we can observe with children for instance.
Having said that, it is becoming more and more obvious that many of us get intimidated by how much of what we do can be replicated and even improved on by artificial systems. One of the many illuminating points made in this book, is how much of what we thought was a testament to human uniqueness, proves to be precisely the type of routine, rule-based activity that a “mindless” algorithm could perform.
The idea behind the human code is that it is supposed to operate in a way similar to the human subconscious. Without relying on any generalized, universal set of explicit rules or axioms, it should continuously generate new algorithms, re-coding itself as it goes about accomplishing specific tasks. It is quite possible and even anticipated, that in the end, neither the AI nor the manufacturers should be able to understand what rule the system decided to adopt at a particular moment. Very much like an artist.
“I would now like to propose a new test: the Lovelace Test. To pass the Lovelace Test, an algorithm must originate a creative work of art such that the process is repeatable (i.e. it isn’t the result of a hardware error) and yet the programmer is unable to explain how the algorithm produced its output. This is what we are challenging the machines to do: to come up with something new, surprising and of value” (Du Sautoy, M., 2019).
This concludes Chapter 1 of the Creativity Code, the title of which, in case you were wondering, was The Lovelace Test. We will continue where we left off with the next post.
REF
Du Sautoy, M. (2019). The Creativity Code: How AI is learning to write, paint and think. HarperCollins UK.
