In 1950, Alan Turing published on Computing machinery and intelligence what would later become known as the “Turing’s test”. The objective of his test was to set a criterion for evaluating the intelligence of a machine, ultimately claiming that if it could have been mistaken for a human being, then it could be considered intelligent. The test and its actual reliability have often been criticized and discussed, and yet, seventy years later, not only have machines far exceeded the conditions set, but the debate has also shifted to trying to understand if they can be defined as creative.
Scheme representing the Turing test: three are the participants (A, B and C). A is a man, B a woman and C any person who has to guess through a series of questions who is the man and who is the woman. B also has to help C in the task, while A has to hinder him. A, however, is replaced by a machine and the aim is still to pass himself off as a human being and make the sexes guess. If C can guess anyway, the machine will have passed the Turing test.
We are looking for a definition
What seems to puzzle most critics and experts in the field – as well as us common people – is the perception that creativity is something ontologically connected to man and his introspection. As a result, it is unlikely for people to associate this capacity with artificial intelligence (AI) unless we want to insinuate – and this question is already at the center of many debates – that the machine can have a consciousness and a proprioception. While not addressing this issue here, there are nonetheless supporters of AI’s creativity according to whom machines can and are creative. The first to propose this point of view is Margaret Boden, an AI expert, a pioneer in the world of cognitive science and a member of that group of scholars working to create a new discipline called “computational creativity”. Since there is no single definition for creativity, it’s worth understanding how Boden stands in relation to it and how she defines it. The expert frames and describes 3 different types of creativity:
As far as an exploratory kind of creativity is concerned, if you take into consideration the fact that machines are already used as a tool that allows man to expand his creative ability, this aspect is already amazing in itself. However, if this is not enough, let’s mention a couple of rather destabilizing examples.
Margaret Boden explains how one can define creativity
The most famous case of possible art created by AI, in fact, is that of AARON, an algorithm created by Harold Cohen that from its first programming does nothing but produce paintings of its own elaboration, continuing even after the death of its creator. This last data, in fact, can make us think that the machine, after the programmer’s death, has acquired independence and – if we add to this the fact that Cohen himself has always said that AARON was actually able to make decisions by itself thanks to the ability to generate random number series – we are further inclined to give it at least a certain degree of autonomy. This case proves to be very fascinating, to the point of remaining at the center of many scholars’ discussion and being an inspiration for other researchers in the artistic-artificial field, but there is another case that may raise even more doubts about the absence of creativity in the machines.
Some works by AARON
In 2016 the company DeepMind challenged the world champion of Go – an ancient Chinese game similar to chess – with its new algorithm called AlphaGo. AlphaGo was born a few years earlier with the aim of understanding to what level the machines could try their hand at an activity in which both intelligence and creativity are important. Expectations were clearly against the machine, with people all over the world believing it would be an almost certain defeat. The battle between the two challengers was hard-fought, but AlphaGo not only managed to win, but indeed managed to tear his opponent to shreds, and, with a single move – the number 37 – to revolutionize the entire game mode. If it cannot be unquestionably claimed that this is a clear example of creativity, it can at least be implied or presumed. The DeepMind team itself was stunned by such an unexpected event. The machine had expanded the established limits of a centuries-old game by shattering them. Now the game of Go has regained notoriety and moves never conceived before – based precisely on move number 37 – are being created and studied.
Explanation of the rules of Go and illustration of the game between AlphaGo and Lee Sedol
In the light of this victory and of the behaviour displayed by the machines described in the examples provided above we can state that Turing’s test has been largely overtaken by them and, indeed, we could ask: “if I saw a picture made by AARON, if I read a book entirely written by a software or if I saw that famous game of Go, and I didn’t know that behind it there is an algorithm, would I consider all of this creative?”
To learn more
- Barzon, Beatrice (2018). «Creative AI: la creatività computazionale nell’arte». Spindox: Digital Soul, 15 gennaio. URL https://www.spindox.it/it/blog/creativita-computazionale-arte/ 2020/06/04
- Du Sautoy, Marcus (2019). Il codice della creatività. Milano: Rizzoli Editore.