For decades we have heard predictions: the world will be taken over by artificial intelligence. In 1957, Herbert Simon predicted that in 10 years the digital computer would become the world chess champion. This only happened in 1996. And despite Marvin Minsky's predictions in 1970 that “in three to eight years we will have machines with the general intelligence of the average person,” we still think of this as science fiction.
The pioneers of artificial intelligence certainly missed the timing, but they were absolutely right: AI is coming. It will be on our televisions and in our cars; he will be our friend and personal assistant; will take on the role of a doctor. There have been more breakthroughs in AI in the past three years than in the last thirty years.
Even tech leaders like Apple have been caught off guard by the rapid advancement of machine learning, the technology that pushes AI forward. At its recent WWDC for Developers, Apple unveiled its AI systems so that independent developers can help it create technologies that rival those already available from Google and Amazon. Apple is falling behind.
In the past, artificial intelligence has used brute force to analyze data and present it in a human-readable form. The programmer clothed intelligence in the form of decision branches and algorithms. Imagine you are trying to build a machine that can play tic-tac-toe. You give her specific rules for what moves she can make and she will follow them. This is how Deep Blue from IBM defeated chess grandmaster Garry Kasparov in 1997 - he was a supercomputer that could calculate all possible moves faster than a chess player.
Modern AI uses machine learning in which you give it examples of previous games and let it learn from those examples. The computer is told what and how to learn and makes its own decisions. What's more, the latest AIs model based on human consciousness itself, using techniques similar to our learning process. It used to take millions of lines of computer code to train a machine to recognize handwriting input. This can now be done with hundreds of lines. You only need a large number of examples for a computer to teach itself.
New programming techniques use neural networks - which are modeled on the basis of the human brain, in which information is processed in layers, and the connections between these layers are strengthened based on what is learned. This process is called deep learning because the number of layers of information processed grows with the speed of computers. Computers learn to recognize images, voices, text - and do human work.
Google search used a technique called PageRank to display results. Using tight, proprietary algorithms, Google analyzed text and links on web pages to determine the most relevant and important ones. Google is now replacing this technique in search engine and most other products with deep learning algorithms. The same technology that beat the world's best Go player. Observing this incredibly complex game, the creators of the method themselves did not understand why the computer makes such decisions.
AI has applications in any field where data is processed and decisions need to be made. Wired editor Kevin Kelly compares AI to electricity: a cheap, reliable, industrial digital intelligence that works just about anywhere. He believes that AI “will bring inert objects to life, much like what electricity did over a hundred years ago. Everything that we once electrified, we will now "reason". The new utilitarian AI will also give us, as individuals, additional capabilities (deepen our memory, accelerate our senses) and collectively as a species. There is nothing that cannot be infused with new, different or interesting using this additional IQ. Business plans for the next 10,000 years are easy to guess: take A and add AI to it, that's all."
Promotional video:
AI will be everywhere very soon. Businesses are pouring AI into their products and helping them analyze the massive amounts of data they collect. Google, Amazon and Apple are working on voice assistants for our homes that will control our lights, order food for us, and schedule our appointments. Robotic assistants like R2-D2 from Star Wars will appear in ten years.
Should we be concerned about the uncontrollability of "general artificial intelligence" spiraling out of control and taking over the world? Yes - but not in the next 15 or 20 years. There are well-founded fears that AI will begin to learn and be able to do more than we do. But big names in the tech world are working to keep it from spiraling out of control and into the hands of the wrong people - like Elon Musk, Stephen Hawking, Ray Kurzweil and others.
AI is coming. And it's definitely worth the wait.
ILYA KHEL