Friday Format: AI — New Enemy or New God



    / photo PROSteve Baker CC

    One of the traditions of the IT-GRAD team is a Friday analysis of what is happening in the world of technology and science. For example, in a series of posts of the “Friday format” we examined the applicability of IaaS in the scientific community ( part 1 and part 2 ).

    Today we decided not to stop there and talk about AI.


    Artificial intelligence has reached the next level of development - Google AlphaGo defeated the Korean grandmaster in the game of Go. Is this an achievement? After all, machines have long dominated professionals in solving intellectual problems - 19 years ago, the Deep Blue computer surpassed Garry Kasparov in chess, and in 2005 the Watson AI system won the Jeopardy! TV show (the prototype of “Your Game”).

    But Go is a more complicated game. If you familiarize yourself with it, it becomes clear that, by calculating the possible moves, you won’t be able to win - to defeat the world champion, the machine needs to learn to “think” creatively, improvise and adapt to the situation. To win go, you needed to create an advanced, truly creative artificial intelligence.

    Computer technology is developing exponentially - the intellectual capabilities of modern computers are compared by experts with the abilities of the mouse, despite the fact that several years ago they were at the level of insects. This fact is alarming, because sooner or later the development of computers will lead to their intellectual superiority over people.

    Stephen Hawking and historian Yuval Harari build negative scenarios for further developments, Ray Kurzweil rubs his hands in anticipation of eternal life in the information field, and writer and entrepreneur Mark Manson urges not to cause a panic. What will “flirting" with the development of technology result in for humanity?

    Platform war


    Computers can not only play board games and participate in quiz games. Ten years ago, people laughed at the low performance of self-driving cars. Today, unmanned vehicles drive not only on a closed highway, but also on a highway loaded with vehicles that people drive.

    A few years ago, facial recognition using software was extremely expensive, and in practice, programs did not identify people very accurately. Such technologies were considered overdeveloped spy toys and were used only by the governments of several countries. Now, Facebook can just pick up and tag your friends from a recent party.

    And hereYou can listen to music written and performed by a neural network. An inexperienced listener is able to attribute authorship, if not Mozart, then to one of the contemporary composers.

    The rapid development of technology is forcing the largest IT companies, such as Amazon, Google, IBM and Microsoft, to take active measures to become the best in the development of AI. As IT industry experts put it, companies have entered the “platform war”.

    A platform in the IT sphere is usually called a software product, on the basis of which other companies write programs, and which ordinary users cannot do without. A company with its own platform guarantees a large profit.

    The most active participant in the development of artificial intelligence platforms is IBM. The Watson division, launched in early 2014 as an IBM subsidiary, is engaged in both software development and the provision of various services. Therefore, the technologies of this project are optimized for specific industries.

    More than 80 thousand developers have already downloaded and tested software products of Watson, a company that cooperates with 500 partners from the industry, including large companies and startups. For example, startup Ross Intelligence has developed a robot assistant assistant based on the Watson AI.

    “This is just the beginning, and in the long run, we want hundreds of millions of people to use Watson as an intelligent, self-service service,” said Watson General Manager David Kenny.

    In 2015, Amazon and Microsoft introduced machine learning capabilities into their cloud platforms - Amazon Web Services and Microsoft Azure. Companies use machine learning algorithms to help customers identify patterns and make predictions based on large amounts of data.

    Microsoft offers 18 machine learning services, including face recognition, text analysis, and recommendations for using the product. At the end of May, as analysts predicted, Microsoft showed even greater prospects in working with AI at a developer conference in San Francisco.

    Google reveals its plans for working with AI technologies in an attempt to attract the attention of developers. In November, the company introduced TensorFlow's open source machine learning technology, which its employees use.

    In May, at a conference in San Francisco, Google boasted a new service for converting speech to text. The company also talked about how it implements an image processing system, which will soon be available for mass use. Google also introduced new tools and tutorials that will simplify the development of applications with elements of machine learning.

    According to IDC, by 2020, the market for applications using machine learning algorithms will reach $ 40 billion. At the same time, 60% of these applications, according to IDC representatives, will be implemented on the platforms of four companies - Amazon, Google, IBM and Microsoft.

    According to the scientist Jeff Dean, who oversees the development of Google’s AI solution, “smart” applications will become commonplace, “and machine learning will affect every industry.” Developing the most advanced AI technology means for the company leadership in the IT industry for years to come.

    “Whoever wins this race will dominate the next stage of the Information Technology Century,” said Pedro Domingos, an expert in machine learning and author of The Master Algorithm, which describes how AI and Big Data technologies will change the world.




    / photo Miranda Granche CC

    What's on Siri's mind


    According to The Guardian, the new book by historian Yuval Harari, whose book is recommended by Bill Gates himself, “Homo Sapiens: A Brief History of the Future” (Homo Sapiens: A Brief History of Tomorrow) contains very gloomy predictions. He admits that artificial intelligence can make some people look like gods and predicts the emergence of a "class of useless people" who do not have time to catch up with computers and their more successful and enterprising relatives in development.

    One day the day will come when we will create a computer that will be “smarter” than any person on Earth. Then computers will begin to use us as a means to solve their problems, and our thoughts, decisions and actions will gradually become unnecessary. Will people then have a reason for living, a reason to wake up in the morning?

    Many key figures of modern science and business also hold pessimistic views. Bill Gates, Stephen Hawking and Elon Musk are just a few of the leading thinkers and scientists who are surprised at how fast AI is evolving and how human is not ready for the consequences of this development.

    When Ilona Mask was asked what we should be afraid of in the near future, he said without hesitation that there are three threats: the first is a global nuclear war, the second is climate change. Before calling the third, he paused. When the journalist asked him what the third threat was, he smiled and said: “In short, I hope that computers will treat us well.”



    Posted by Ilona Mask on Twitter: “I hope we are not just a mechanism for launching digital superintelligence. Unfortunately, this outcome is becoming more likely. ”

    Perhaps the most open and respected technology armageddonist is the Swedish philosopher Nick Bostrom. One of the fears of Bostrom and its supporters is the emergence of uncontrolled self-improving technology - a machine that is so “smart” that it can make itself, or new versions of itself, even smarter without human intervention.

    If it reaches a level that exceeds human intelligence, then, according to the law of accelerating returns, the launch of the exponential development of technology is just a matter of time, and we will not be able to stop this process. Bostrom makes an important remark here: the creation of a superintelligence that is smarter than us can destroy the entire human race.

    Some of them are looking forward to the revolt of the machines, because they believe that their superpowers to solve problems will surpass ours so much that life will become incredibly happy and free from problems. All diseases will be cured. Poverty, famine, war and environmental problems will disappear. We will all have endless free time, and in a certain scenario, as some believe, machines will make us immortal.

    Ray Kurzweil, for example, argues that technology will not only improve our lives, but can save us and, possibly, guarantee us a place in the universe for an indefinite period. He believes that future technologies - for example, nanorobots - will be able to restore our cells, reverse the aging process and remove excess fats and sugar from the body so that we eat whatever we want. In the event that our physical bodies still do not exist forever, according to Kurzweil, we will have the opportunity to unload our consciousness into a "cloud" and live in a virtual world even after our bodies disappear.

    Kurzweil and his supporters believe that technology has no reason to harm humanity, as it is not just created by us, but is becoming more and more part of ourselves. Kurzweil supporters suggest that the moment will come when the border between biology and technology is barely noticeable. In this case, any technology that causes harm to humans, at the same time, will harm itself, and any self-destructive technology will not last long. Therefore, it will "die out" as quickly as a harmful mutation that is excluded from the gene pool.

    Other adherents of technological utopia suggest that artificial superintelligence could answer questions that a person is not even able to understand. Thanks to this, the standard of living in the world will grow exponentially. In addition, machines will not just invent more advanced devices: they will increasingly come up with more efficient methods for creating these devices, so almost all living things can reap the benefits of their labors.

    It turns out that the proponents of the development of AI deify him, expect only mercy from him and are unlikely to realize that a terrible judgment may await them instead. While Utopians and Armageddonists argue, entrepreneur and writer Mark Manson looks at the problem indifferently, believing that computers will have a high level of morality and will try to make people's lives sweet and carefree.

    And if they need to destroy people — they will choose the path of least resistance — they will allow humanity to self-destruct, only by directing their actions in the “right” direction. What does he think of enslaving computers? Overly happy slaves never rebel.

    In any case, the statements of optimists, pessimists, indifferents and other representatives of the human race are just guesses. What machines plan to do with humanity, surpassing it in development, is known only to themselves. And they are unlikely to reveal their plans to people. If you ask, for example, Siri, whether robots will take over the world, she will answer: "It does not matter what I think." And when asked about the emergence of a class of useless people, you will find an evasive answer: “And this is after all that I have done for you?”

    Only registered users can participate in the survey. Please come in.

    Forecasts are a thankless task, but ...

    • 51% For the future of AI, it will become available by analogy with computer technology, everything is fine 97
    • 16.3% The best is the enemy of the good, but leading companies should not close AI developments to the public 31
    • 11% Dreams of real AI are nothing more than utopia, you can relax 21
    • 21.5% Which AI? We have already understood what society needs. And this is Pokemon GO 41

    Also popular now: