 Direct Translation via Google Translate. Edited.
[REGNUM] There is nothing like a person - good or evil - behind artificial intelligence and is unlikely to appear. In this regard, AI should not be the subject of either excessive fears or unreasonable hopes. But there are many real dangers associated with it.
Alexa, play some Sig Kowalski polka music.
Artificial intelligence GPT-4, like other neural networks, is now in the center of interest, some enthusiastic, some scared.
Robots are beginning to do what until recently was considered the exclusive prerogative of man - to compose meaningful texts. Articles, poems, sermons, even peace treaties. They can draw pictures on the instructions of users much faster than professional illustrators.
Now anyone can ask a neural network a question and get a very meaningful answer. What's more, there are plenty of phone apps that let you — if you're feeling sad — chat with a patient and always friendly companion.
Many say that we are experiencing "the biggest technological revolution since the invention of the wheel." Some share panicky forecasts - neural networks will at least bring down the labor market, leaving a lot of people of “creative professions” out of work, or even take over the planet, doing away with people.
Others, on the contrary, are full of actually religious hopes - artificial intelligence will solve all our problems.
Wise computers will develop even more wise ones without the help of people, they will give rise to an intellect so powerful that it will save us from all misfortunes, and even death itself. And what zealous adherents of technocults call "singularity" will appear - a good transformation of the entire universe by the power of an artificial mind, which, having learned all the secrets of nature, will even find a way to resurrect the dead.
However, these fears and these hopes have little to do with reality.
In fact, of course, artificial intelligence is fraught with great opportunities and serious dangers, but not in the way that people usually imagine.
When computers do something that used to be considered exclusive to humans—talking or drawing pictures—there is a strong temptation to assume that there is someone like us inside them, with will and intelligence. Someone who can make decisions or hatch designs, good or evil.
But it is not.
Sometimes, for example, a neural network is accused of “racism” due to the fact that the pictures it draws are almost all white. But this is by no means a sign that a computer program has a personality and that personality somehow has a grudge against black people.
Everything is much easier.
The network is trained on an array of images on the Internet, and if you go to any resource where photos are uploaded, for example, on flickr.com, you will see that most of the people in the photos are white. For the obvious reason that there are much more opportunities to buy high-quality cameras and travel around the world with them among residents of Europe and prosperous regions of North America.
The same effect leads to the fact that neural networks are worse at recognizing the faces of minorities - simply because there are fewer of them on the Internet.
But we tend to fall into the understandable illusion of ascribing preferences or even prejudices to computer algorithms. In reality, programs cannot show "love" or "hate". They can't even think, and it's not a matter of computer power.
Back in the 19th century, the German philosopher Franz Brentano drew attention to such a property of human thought as "intentionality", that is, direction, "clearness". Thoughts always have their own object, they always turn to something, they always have an element of personal, subjective choice. You are thinking about this subject and not about another.
The computational process is completely devoid of this quality - it is not "about something", it simply processes the data according to the algorithm.
If, for example, you enter a text about the work of early Mandelstam into the window of an electronic translator, the program (unlike a human translator) does not “understand” a single word in it. She simply performs a substitution depending on the array of translations on which she was trained.
If a neural network gives you a snippet from an article, this does not mean that it understood its content and retell it in its own words, as a person would do. This means that its algorithms have processed the text, having been trained on many examples of how people do it.
As the philosopher Daniel Dennett put it:
"Artificial intelligence in its current manifestation is a parasite on human intelligence. It devours, rather indiscriminately, what was produced by the human creators and squeezes out the patterns it finds there, including some of our worst habits. These machines do not (yet anyway) have goals or strategies or abilities for self-criticism or innovation that would enable them to go beyond their databases and think about their own thinking and their own goals."
That is, nothing resembling a person - good or evil - is behind artificial intelligence and is unlikely to appear. In this regard, AI should not be the subject of either excessive fears or unreasonable hopes.
But there are many real dangers associated with it. To mention only the most obvious: although the program is not a person and cannot be your friend or confessor, it can successfully imitate both.
A pleasant electronic voice does not get annoyed, does not condemn, does not disagree with you, it always has time for you. And the temptation to replace communication with people, often heavy, uncomfortable and prickly, with communication with AI can be almost irresistible.
Social media has already led to the formation of "echo chambers" when we find people who agree with us (even when we are clearly wrong) and reinforce our prejudices and delusions.
AI will allow people to lock themselves away from their neighbors in a self-imposed prison of gentle voices, none of which will be human.
This can destroy not only communication, but the very ability to communicate. Not by the evil will of the algorithms - they have no will. But because such is already the unfortunate state of fallen man - to turn everything to his own harm. Including artificial intelligence.
April 19, 2023
Sergey Khudiev
|
