THE ALGORITHM REVOLUTION

The advances in computing are incredible. And they won't stop. One of the latest and most outstanding advances is Google's API. The algorithm created by Google is already capable of distinguishing violent or offensive content, extracting texts, differentiating colours... but it does not quite understand art. Several users have put it to the test and this is what they have learned.

From René Magritte's famous piece "Ceci n'est pas une pipe" ("This is not a pipe"), Google sees just that: a pipe. Something we might even consider logical. But for Georges Braque's "Woman with mandolin", the algorithm thinks it sees archaeological remains. Why is that? Matt Watts, head of data strategy and chief technology officer at NetApp, points out in a article that algorithms "lack the built-in intelligence to understand the finer aspects of art". In other words: the algorithm lacks imagination.

Although in "Woman with Mandolin" API Vision is unable to recognise either the woman or the instrument, he does manage to classify it as art. The same is true for other works by Kandinsky. The artist's watercolours are identified as "art", "illustration" and "graphic design".

There are times when algorithms do not act as we expect them to. Nor as they are programmed. Many of them are "rebelling" against their creators. Although, we don't have to be so alarmist. We are still a long way from an apocalypse where washing machines and mobile phones kill us.

At the moment, the different algorithms are being tested in different situations. The aim? So that they learn to cope with different situations. But the ones who are learning the most are their creators. Here are a few cases in which the algorithms surprised, and very much so, creators and users.

One of these examples The events that occurred during the development of NERO were anecdotal. This had been presented as the first videogame where characters that were not the player could evolve as the game evolved. The idea was for players to train their own robots and one of the tests was to teach them to walk between walls. While a person would have gone around the wall, the NERO robots climbed up and down the other side.

Another took place during a simulation to study the evolution of communication. Robots were created with wheels and blue lights, which they had to use to dialogue with their companions. The main task of these robots was to find food and avoid poison. If they succeeded, they were rewarded. After several generations of robots, they adopted two behaviours. Some would alert others when they found food by using lights. Others tricked their fellows and turned on their lights when they found poison. In this way, they were able to eliminate their rivals.

A "sick" intelligence has even been created. It was created by MIT and named Norman, the first artificial psychopath. Norman was the first artificial psychopath. Why do we say this? Because, while a "normal" system sees a group of birds high up in a tree branch, Norman sees an electrocuted man.

But, as we said, we should not be paranoid.