Work on artificial intelligence (AI) began before most people began to suspect that such a thing was possible. In the same way, people previously doubted that it was possible to fly into space, and even more so to the moon. But it happened. Work on the creation of machine intelligence is in full swing and AI is improving in proportion to the increase in computer performance. Therefore, the creation of a full-fledged artificial intelligence, approximate or, rather, superior to human intelligence, is only a matter of time.
Almost every month we receive news about another outstanding achievement of neural networks. This is due to the tremendous interest from investors and top scientists. But the actions of AI are mostly different from those of humans, and the situation is unlikely to change anytime soon.
Purpose of Artificial Intelligence
AI systems, like any other technology, are created in order to make life easier for people. However, this situation also carries threats: smart machines will completely take over some professions, and this process has already begun. Neural networks are now used by banks, social networks and trading exchanges. It's all about the speed of their learning and subsequent work. They make hundreds of times more trades per day than any experienced person.
And, unfortunately, this is not the only reason to be wary of artificial intelligence. Influential people such as Stephen Hawking, Bill Gates and Elon Musk fear that intelligent machines will become the dominant species on our planet. They will become smarter and faster than a person and will not want to serve him, and instead will begin to use people in their own interests, say the above authorities.
Moreover, machines will not even use brute force: they can turn people against each other when they subjugate the media and start spreading false news.
It is too early to judge how much humanity can prevent such a catastrophe, but one thing can be said – it is quite real. And a number of already known artificial intelligences are proof of this.
to the content
The most famous AI systems
The list of the most popular AIs is constantly growing, so the list below is just the beginning.
to the content
Deep blue
Defeated World Chess Champion Garry Kasparov. It happened back in 1997. Deep Blue (translated as dark blue) defeated the grandmaster with a small gap in games. It was already the second match (a man won the first). But this outcome did not please either the chess player or the computer scientist, and after that Kasparov announced an unfair outcome and high psychological pressure due to the special attention to the upcoming match. Therefore, the project was closed and Deep Blue no longer appeared in tournaments. A play dedicated to the last match was released in 2013, and in the same year Disney bought the rights to film it.
to the content
Watson
A supercomputer equipped with a question-and-answer artificial intelligence system. He became the winner in the game Jeopardy !, which analogues are popular in many countries of the world. His opponents were 2 of the best players: Brad Rutter, who received the largest win in the history of a TV show, and Ken Jennings, who broke the record for the longest winning streak.
to the content
Libratus
Libratus is famous for beating the top 4 poker players in early 2017. At the end of the 20-day match, the players lost $ 1.8 million to the computer.
Initially, the Libratus program only had an idea of the game's combinations and understood which one was better. But the developers instructed her to play with herself to learn from her own mistakes. When the machine played several trillions of hands, it began to operate as efficiently as possible to maximize profits. The most interesting thing about this is that the Libratus AI learned to bluff on its own.
From the very beginning, the computer started winning, but on day 6 the players managed to close the gap. After that, artificial intelligence began to play even better and rapidly increased its advantage.
The developers claim that this creation can be used not only for playing poker, but also for other situations with incompletely open information: risk assessment in bidding and negotiations, and others.
In general, the speed of self-learning of AI systems is thousands of times higher than that of a human. So far, artificial intelligence is not applicable everywhere, and cannot independently switch from one profession to another, but the question remains whether this will last.