Defense science agency DARPA is preparing to launch nearly four-year project on the development of artificial intelligence, which will be able to educate them and improve them. In the new project is not about modeling the human brain and run the model on a computer.
This way in the near future, many scientists believe a dead end because of the exceptional complexity of the human brain and the little-studied principles of our mind. DARPA plans to go the other way: to build a machine that will learn with the help of algorithms probabilistic programming that is sift through massive databases and choose the best solutions to problems. In the process, artificial intelligence will learn and after a while can easily solve simple problems. It is with such projects in science fiction novels begins Rise of the Machines and the loss of humanity. DARPA experts offer commercial campaigns for 46 months to develop new technologies for programming probabilistic learning machines, abbreviated PPAML. PPAML military technologies more effectively help to solve many analytical problems, which today require huge human resources. In particular, learning machines will be extremely useful in the conduct of intelligence, surveillance, speech recognition, driving a car, screening information in search of valuable data and more. Hardware artificial intelligence can be diverse: supercomputers based on multi-core processors, the network of conventional PCs and cloud networks. It is difficult to say how realistic in the next few years, the creation of a complex system of artificial intelligence. Currently, there are many automated analysis tools to facilitate the work of people, but so far none of them is capable of real self-learning. As long as the supreme achievement of technology to the database remains essentially collating archival images with images from the camera. Therefore, in the DARPA stress that PPAML algorithms are much more complicated, in particular, they will be able to integrate the information of different types, independently formulate the request and based on the data to formulate a solution.