AI is a great unknown: for many it is still something related to science fiction, although it is increasingly applied in our daily lives
Almost all the things that we don’t understand and that fascinate us are also a little scary. That’s the thing with artificial intelligence (AI). For many, our first contacts with this term were linked to science fiction: to movies, stories and comics where a ‘machine’ endowed with this ‘virtue’ ended up rebelling against humans and causing havoc. And although that idea of artificial intelligence – very childish – has already been overcome, some trace remains in the popular imagination. «Everyone has heard of it, although they do not fully understand what it consists of. Sometimes it is presented as a magical solution to a multitude of problems, while in others it is portrayed as an apocalyptic threat”, indicates Javier G. Algarra, academic director of the Engineering and Sciences area at the U-tad University Center. This Friday, Artificial Intelligence Day, this expert dismantles the six most widespread myths about this technology that is already here.
It will destroy many jobs
It is a widespread myth and a very, very real fear, since in many areas AI is replacing human work. “AI will end jobs, but only those with low creativity,” says G. Algarra. The good part of all this is that, in return, it will cause others to appear in which more trained personnel are needed. In other words, rather than destroy employment, it will transform it.
It is applicable to almost everything
It has multiple applications, yes, but it is not the universal panacea for everything. “It still lacks creativity,” says the expert. Thus, the AI can generate very realistic synthetic images, imitate the way of painting of the great masters or even write apparently perfect press articles (perhaps this one?), but why artificial intelligence to ‘achieve’ do all these things needs to ‘learn’ from humans, who are the ones who load the system with material. In which areas does AI work best? “We all remember when in 1997 a machine defeated the world chess champion. It also works extraordinarily well at recognizing images: that’s the basis for machine vision in a self-driving car or automated medical diagnostic systems. Likewise, it is useful in the recognition of sounds and in recent years a lot of progress has been made in the automatic translation of language”, enumerates G.Algarra.
Makes machines think and act like humans
“It’s not like that,” says the specialist bluntly. As he explains, the way a microprocessor works is very different from that of our brain: AI is not a carbon copy of our ability to reason, creativity or affectivity… «And we are still a long way from being able to emulate them, although some applications they can show apparently human behaviors”, advances G.Algarra.
It is unfair, insensitive and makes mistakes
Detractors of artificial intelligence seize on the argument and some notorious errors to say that AI can be racist, sexist… and, ultimately, be unfair or insensitive if it is applied, for example, to the selection of personnel for jobs , to the recognition of suspects in investigations… Come on, you can’t trust them like a human (as if humans weren’t loaded with prejudices). What does the expert think? “AI applications produce undesirable results if they have been trained on biased data,” he says. That is, if it has ‘learned’ badly. And you have to understand learning for what it is: that this application has been endowed (the humanitis) with multiple comparators connected in a chain. “For example, if we train a network to recognize images of dogs and then present it with a photo of a cupcake, it can determine that it is a chihuahua, as has already happened with a very famous example,” recalls G. Algarra. This can be a nightmare if it is used to recognize wanted people and in the examples there is a percentage of images of a racial or physiognomic type much higher than that of the general population. AI lacks a moral conscience.”
Use artificial neurons to learn
Oh, people make a mess. The artificial neurons of AI is a metaphor to explain to people how the learning of this technology works, only that perhaps it has served to confuse people more than anything else. When it is said that AI is based on neural networks “we are talking about a combination of simple mathematical functions loosely inspired by the electrical behavior of a neuron,” says the expert. Inspired, not neurons per se!
Will ‘machines’ rule the world?
In ‘2001’ the HAL 9000 computer rebelled against the crew. In ‘Terminator’ the robots took control of the planet after the super computer Skynet triggered a nuclear war… scary, right? Could it come to pass? Isn’t that emancipation of the ‘machines’ a real risk? “This dystopian vision has its origin in science fiction,” explains the U-tad expert in Engineering and Sciences. Artificial intelligence is not going to kill us of its own volition, since it doesn’t have to. The reality is that human beings do not need the help of machines to destroy our common home, as we demonstrate daily.
#true #topics #artificial #intelligence