Models of artificial intelligence, developed by Alibaba, and Microsoft, received the highest results in the test at Stanford University on reading and understanding texts.
The test SQuAD is a collection of more than 100,000 questions submitted texts. Mostly clippings of Wikipedia articles. The task of the dealer — find in the text information relevant to the question. The best result among men currently stands at 82 304 points, while the AI from Alibaba scored 82 440 points, and Microsoft — 82 650.
Independently formulate answers machines have not yet learned, but to highlight the most basic know how already very good. Chief researcher in natural language processing at the Institute Alibaba suggests that further development of this technology can be applied in various applications. For example, as a technical support or online with online medical consultation.