Why experimental art matters: Looking at art from an A.I perspective.

“We, writers, painters, sculptors, architects and passionate devotees of the hitherto untouched beauty of Paris, protest with all our strength, with all our indignation in the name of slighted French taste, against the erection … of this useless and monstrous Eiffel Tower …                                                                                                                                                                  “Artists against the Eiffel Tower” petition,14 February 1887

Did you know that the Eiffel Tower, the symbol of Paris  the most visited city globally with 80 million tourists per year , was heavily confronted and fought against,by the contemporary artists of the time? When I first got this information I was quite impressed and stunned, but then again, it’s easy to judge after the fact.

Recently, I visited Tate Modern and I didn’t understand almost any of the exhibits. They looked to me like a  pretentious, self-praising, pompous waste of time.We all, at some point in our lives, came in contact with weird and experimental art. Whether it was the Holy Mountain or whether it was something like that:

the-weeping-woman

Let’s be honest, we have all wondered “why is this called art?” or “what is this good for?”. I am not going to get into the what-is-art argument, but I am here to tell you this weird/experimental/underground art does serve a purpose: It keeps the rest of the Art healthy preventing it from suffering a death by stagnation and I am going to use A.I in order to prove that: They are the exploration nodes of the human intellect.

Exploration vs Exploitation in A.I

There are a lot of A.I algorithms designed to solve vastly different problems, but a running theme in all of them ( or most of them, I haven’t seen all of them to be certain), is the choice between exploration and exploitation. Exploitation is when the A.I is choosing to do something, because it believes it’s the best thing to do. It’s the “I am doing this because, to the best of my knowledge, it’s the best move to make” mentality. Exploration is when the A.I is choosing to do something randomly, because ,who knows?, maybe something will come out of that. It’s the “Let’s do this random thing and see what happens” mentality.

Here is a simple example:Let’s say that you have some free time in your afternoon. You have two choices either do what you always do, which is  watch Netflix, or try something new ( you don’t know what, but it sounds exciting!). The first option is the exploitation ( you are basing your decision based on what you know) and the second one is the exploration.

exploration_exploitation

To get more technical the simplest method to implement an exploration vs exploitation is the \epsilon-greedy method, where  the agent chooses the action that it believes has the best long-term effect with probability1-\epsilon, and it chooses an action uniformly at random, otherwise.

Now, the natural response is, “wait I thought that A.I was supposed to be all about thinking smart, what do you mean, they make the A.I random things?”.

Artificial Intelligence scientists have realized that only using your current knowledge to perform actions is not the best way to solve a problem. This will lead you to be stuck in a small area of the solution space. A healthy system needs to have a small element of randomness. Just like in our own example: If you only decide to spend your evenings Netflix, you will have an okay life, but you may be missing opportunities a lot of opportunities to have a great life. For example maybe you should have accepted that invitation to try an escape room, or to go out with some friends: Maybe this would lead you to discover a new passion in your life, or meet someone exciting and interesting.

In other words occasionally you have to try something new.

The tricky part in A.I is to find the balance between exploration and exploitation. You don’t want to get stuck in the same routine, but don’t want to force yourself to do something new every day either, your everyday life would lose consistency and meaning. It’s the same with A.I . Too much exploration and the A.I agent will just end up doing random insequential things. Too much exploitation and the A.I will be stuck in a local maximum. It’s the balance that makes an agent able to discover the best solution available for a problem.

Exploration vs Exploitation in Sciences

The same principle applies  for art and human thought. Quit often I hear the argument, “why do I care about CERN or the God Particle?”. I am pretty sure some wise-ass told Newton in England “why is is useful to know how fast the apple is falling if I get hit anyway?”. The real answer is : You don’t know. However, just because you don’t know the immediate results or implications doesn’t mean that it’s not worth exploring. Einstein’s theory of relativity, when published, seemed an obscure hypothesis concerning only Physicians closed in their libraries for far too long, completely disconnected from our everyday lives. Yet, today we use this theory every day in our phone’s GPS systems. This is why it’s important to have research in Universities and not only funded by companies. Whereas, company funded research is important it will tend to be a little more short-sighted and geared towards profit. It will not venture into the unknown or try to think outside of the box. Again, we have exploitation vs exploration concept.

Exploration vs Exploitation in Art

So, back to art, next time you see a very weird art-piece remember 9 out of 10 times it will be just something utterly meaningless, self-entitled piece lacking any substance. Anyone having visited Tate Modern in London can attest to that.  These 9 pieces though serve as the incubator for that tenth little gem which will become a catalyst to progress a whole medium and for us to see things a little bit differently. They are the exploration factor of our system and they prevent the rest of the system from becoming too narrow-minded and parochial. They prohibit it from getting stuck in a local maximum. Do they sometimes overdo it? Yeah, most of the times, but yet again maybe it’s alright for Art to do that, since it always represented the less orthodox and logical parts of our thinking process.

This is why ,although art almost never has practical applications it is equally important it does not  become stagnant and lifeless. Art affects human thought and it is a source inspiration and influence for the rest of us. The Catholic Church used art in the form of the Baroque theme to fight the Protestant reform.   If artists were not experimenting we would still be having only Greek Drama for theater and Church music on the radio. Our paintings would still be about red hands on caves.

So, next time we see a weird piece of art remember: It’s the exploration price we have to pay in order to  remain vibrant and alive , the ticket toll in order to ameliorate the standards the medium as a whole.