Over the past month or two, we have turned to the topic of artificial intelligence more than once or twice. The reason for this, of course, was not only growing curiosity, but also the hype raised by Musk and Hawking.
Why the hype? Because it will be a long time before artificial intelligence will go beyond the semantic field of "big data" and "clouds". The problem is that people exaggerate the capabilities of artificial intelligence at the moment and continue to misinform the public.
By and large, people view artificial intelligence as a mythical pink unicorn of technology: elusive, powerful, dangerous, mysterious, and well-established. And while there is a lot of controversy in the scientific community now, let's still find out what artificial intelligence is not.
First of all, don't be afraid of artificial intelligence. It is not a living being, like SkyNet, or an evil red light, like HAL. In essence, AI is nothing more than a computer program, smart enough to perform tasks that would normally require qualitative analysis by a human operator. This is not an ubiquitous war machine.
Second, AI is not alive. While AI is capable of performing human tasks, artificial intelligence is not as alive as we are. He is not capable of true creativity, emotions, desires, in addition to what we program, well, or what he learns based on the environment. Unlike science fiction (mostly literature), in reality, AI has no desire to create a family, copy itself or love.
Third, AI is generally not that ambitious. Indeed, in a limited context, AI can think like we do and challenge itself. But its general purpose and meaning of existence is determined exclusively by us from the very first word.
As with any program or technology, we define the role of AI in our society. Be sure that the plans of the creators of AI do not have the ability and intention to create an AI that will enslave humanity and control us like some god from a machine.
Promotional video:
Moreover, AI is not represented as a single whole. Computer programs, even those based on artificial intelligence, are good specialists in a narrow, but not a wide, profile. The likely scenario for achieving artificial intelligence in our life is through a network of subroutines that will control computer vision, speech, adaptation (machine learning), movement (robots), etc. AI is not he or she, it is "they".
Finally, AI, like all computer programs, is controlled by humans. Of course, AI can be designed with malice and armed with nuclear or biological weapons, but this is not the fault of science, but the fault of humans.
While Elon Musk is a kind of hero for many of us, and indeed the genius of our time, his recent remarks about artificial intelligence do not shine with foresight. He said that AI is "more dangerous than nuclear weapons" and that we are "trying to summon artificial intelligence demons." Perhaps Musk fell asleep watching The Terminator.
Meanwhile, companies like IBM, Google and Apple are developing next-generation AI applications using small chunks of specialized code to replace the human element in tedious, dangerous or time-consuming work. These are very specialized, kind of "tunnel" programs that only improve our society and which definitely should not be feared.
In any case, we are decades away from the AI "singularity" that Musk is talking about, and the end result will be closely related to our intentions. If history teaches us anything, it is that when the time comes, we need to fear not artificial intelligence, but the people behind it.