The year that has just ended was marked by a series of lines of force. The climatic consciousness was one, although for now it seems to only concern ordinary citizens. Judging by the decisions that have been taken, a good part of the ruling class must already have a ticket to go to live in a paradisaco exoplanet. In fact, everything that scientists, activists and some communicators have been warning for decades is still falling apart.
Pretty important fact 1:
they do not exist, nor at the moment there could be the slightest possibility of moving to another planet.
Earth is the only place we have in the Universe.
Another of the lines of force was that of artificial intelligence (AI), with which something similar to the weather occurs. He has reached the public discourse and is in matters as everyday as the movies and series that Netflix recommends or the routes suggested by Waze, but it almost does not appear on the political agenda. Although it is clear that it is changing and will change the economic and labor rules of the game dramatically, political discourse sounds exactly the same as in 1960.
Pretty important fact 2: We are no longer in 1960. In fact, almost nothing we do in 1960 is done in the same way, and many of the tasks at that time have disappeared completely. And a huge number of trades and skills that are key today do not exist at that time (nor could they have been anticipated).
Myths and miracles
Something else happens with AI, and although it is quite typical of any new technology, it is still very irritating. I am referring to the crowd that talks about the subject without even having the slightest idea of what this family of sciences and techniques works for. (For example, how does a neural network operate to recognize faces, letters or numbers?)
The consequence is an enormity of folly and gigantic cumulonimbuses of smoke. Last year, for example, they created algorithms that were able to paint a picture. Then, we, the humans – that is why we are human – jumped from "painting a picture" to "making art" in a Santiago. Let's make it easy: painting a picture is not always art. In fact, the subject of art is so complex that not painting a picture could be interpreted as an art form. But the portrait of Count Belamy led us to think that now machines can create artistic works.
Anyway, to those who fell into that sophistry I recommend – at least, to begin with – the booklet
Art and poetryby Martin Heidegger. They will discover that determining what art is much more difficult than we anticipate. If possible.
But, hell, I forgot, we are in the era of intellectual fast food and any text that lasts more than one paragraph or a video of more than a minute and a half is indigestible. Rigorous definitions? God, how tedious!
Music for your ears
Well, in 2019 we will not miss the opportunity to have the
New Tontera IA of the O. A few days ago we learned that a group of musicologists and programmers have joined forces
to complete the tenth symphony of my beloved Ludwig van. This, they say, is because they will celebrate 250 years of the birth of the great Beethoven in December. As a tribute, I do not see the grace, you will see.
The headline sounded good, the idea was catchy like a jingle, and we returned to shield the true artificial intelligence with a crazy armor.
For starters, Beethoven's tenth symphony is not an unfinished work. If it were, the experiment will be delirious anyway, like trying to prove right away. But it turns out that from that work there have been a few measures and
It is considered, therefore, hypothetical, not inconclusive. The Sinfona N 8 by Franz Schubert is an unfinished work, because there were at least two movements and a third survived, although the score is only for piano.
Let's clean up the matter. Now, in addition to all the miracles attributed to AI, we intend to reconstruct a hypothetical work. In that case, and apart from the fact that this type of approach conceals the true wonders that AI can do, it leads to an absurdity. I explain.
Suppose Ludwig had left no notes on the tenth. Suppose you only chatted with a friend. Could an algorithm recreate it? The answer is a categorical no. But artificial intelligence is so clever (or so artificial) that with a few measures, yes, you can give us several versions of the symphony Beethoven had in mind. In conclusion: the artist did not create anything. His works are something that AI can deduce on the basis of a few measures.
It is worth remembering here that these loose fragments of a hypothetical tenth symphony were written by Beethoven at the same time in which he composes his extraordinary (and too advanced for his time)
Last String Quartets. I say it because those quartets were anything but predictable.
But there is another reason why the mere claim that AI could complete the tenth symphony is absurd: we can never verify whether it is right or not. The musicologists will give the OK to some version, and, nobility obliges, they will never register the artificial tenth as part of the artist's work. But we cannot travel to the past and ask Ludwig if the algorithms did what he had in mind. And if we could, I imagine Beethoven answering the most obvious: "I have no idea, I still didn't write it."
In this absurdity lies the fallacy of the whole issue. It is not important that the protagonists are algorithms or musicologists (Barry Cooper already tried, by the way). The tenth never existed, and as Beethoven passed away in 1827, it will never exist. The idea is not that difficult to understand and is directly related to a series of definitions. A symphony is a work composed of a musician. To compose a work, the musician must be (at least) alive. With the tenth, neither of these two conditions exist. Beethoven has been dead for 192 years and nine months, and the symphons are composed, not supposed, deduced, calculated, collided, inferred or extrapolated.
In total, what seems like a powerful headline, ends up repeating something we have known for a while. Machines can create combinations of sounds that sometimes sound like music and sometimes not, through fractals, cellular automatons or algorithms. In the case of the Beethoven project, of course, they incorporate all the artist's work and replicate his style. Thanks to these techniques, they are even able to do beautiful things, like Count Belamy's painting. I am even willing to grant that machines can create some form of art, the day they feel like doing it. The desire, the pulse or whatever moves an artist to create his work.
The reason they are unable to create art (no matter if it is to predict a hypothetical or unfinished work) is that they do not intend to do so. They can create decorative objects or something like that. Similar to art. But not artistic works. Because they don't care.
Of course, all my argument could be demolished by changing the definition of what art is. We are here a problem. Art is something so complex and unattainable that when we try to define it, the definition is altered by the art itself. We are sure of very few things when trying to catch the nature of the artistic work. For example, we know that a motley and beautiful rock is not art; It is the result of some geological phenomenon. Well, machines have no more desire to create art than geological phenomena. The desire, the pulse. Or whatever.
. Beethoven in the labyrinth of artificial intelligence – LA NACION