2016, Oxford University researchers published a report that the chances of writers were replaced by artificial intelligence is very small, far less than some of the possibilities of physical labor and repetitive substituted. But in recent years, the rapid development of artificial intelligence, writing, showing broad prospects for development.
Writing rapid development of artificial intelligence
Some finance, sports news as well as advertising and mail we now see, a lot is written by the program's artificial intelligence, robots such as the Associated Press editor and compiled financial report, Tencent automatic writing program has been able to write sports articles and Finance Wait. In addition to writing these non-fiction writing, artificial intelligence is also poetry, novels, plays and other cultural and sports have achieved some results. As early as 1962, the United States developed the software automatically poetry, his poems published in the "Horizon" magazine. In 1998, the American Academy 纽约伦斯勒 "mind and machine laboratory" of Bruns Ward, who developed a novel writing program "Brutus", only 15 seconds to create a novel. And in 2013 by the American software developer Da Lvsi. Ka Semi-sponsored NaNoGenMo (National Novel Month) have become famous writing machine summit participants to write computer code required to write stories. In addition, the creation of artificial intelligence can play in 2016, according to the movie AI program writing the script of the film "The sun rises" also named a finalist for an international film festival.
In recent years, writing artificial intelligence technology has also been rapid development. Microsoft Research Asia, the development of artificial intelligence program "wheatgrass", after studying thousands of poems 519 Chinese modern poets, wrote and published a collection of poems "Sun lost windows." Poetry writing program Tsinghua University developed the "Nine Songs", it is possible to generate different style of poetry set poem, recent poetry, acrostic, modern poetry and so on. I currently also working with the Institute of World Literature, Peking University, research and development of artificial intelligence biography writing program.
AI writing has experienced 50 years of development, has been involved in news, advertising, financial statements and other non-fiction writing and poetry, novels, plays and other fiction writing. Therefore, the "artificial intelligence literature" in a way has become a reality, it refers to the "AI programs as the main body of writing, writing mimic human behavior and mechanism automatically or semi-automatically generated literary works." Artificial Intelligence and writing some networks automatic writing "Artifact" is not the same, the former can generate new text, but will not come online search sentence together into paragraphs. However, it should be pointed out that, at present, AI is still writing human need to set up and supervised, independent writing artificial intelligence in the true sense is not yet implemented.
Artificial intelligence technology should be improved Dalian gynecological where good mobile.0411fuke.com
Early writing artificial intelligence technology is mainly structuralism, is a top-down model, that is, pre-placement agent for the writing module, and for the establishment of specialized databases, this approach also known as expert systems. Such as "Brutus" system, the story is broken down into multiple dimensions, including plot, characters, literary themes, including the writing style, the advantage that does not require a lot of data, processes and clearly smaller error rate; it disadvantages are also obvious, that is, a single style, people constantly need to set up a new program.
Popular machine learning and the combination of big data, it is a function of adhering to the doctrine top-down model. This approach requires the machine to enter large amounts of text data, allow the machine to master the law texts and learning through self-identification, which is Microsoft "wheatgrass" mystery writing poetry. This depth learning techniques based on artificial neural networks, artificial intelligence makes possible the creation of an independent to a certain extent, people have to do is pre-label and input text data, but this technology requires a large amount of text data, and its label and classification of work is extremely heavy. AI still needs a lot of labor, not intelligence.
Technical experts try to integrate these two effective ways, but both still in the early fusion state: pre-set portion of the frame, and then enter the part of relatively simple data, thus reducing human involvement and supervision. As Baidu automatic writing program, which core processes "auto writing" is divided into "document planning", "micro-planning" and "surface to achieve" modules, and to input data from each of the different modules. After together, set the initial program and the future with true artificial intelligence agent for the integration of people in a certain amount of input data, the machine will be able to do unsupervised learning, classification and grab their own data, generate different styles of work.
In addition, we also see that the current artificial intelligence, writing intention is not to create works of art for people's entertainment, that initially it was a machine natural language understanding, visual recognition and affective computing and other technical issues. Such as Google to optimize the interaction of APP and its people, improving Google products on the user's response, accuracy, training artificial intelligence engine to "read" the 2865 romantic novels, this training enables the machine to better understand the subtleties of human language place. Of course, the "product" is artificial intelligence to learn to write fiction. It also shows, art and technology can be mutually reinforcing. In addition, artificial intelligence by learning human literature and art forms, not only can recognize and understand human emotions, it is also possible to learn the human metaphor - association ability, and this ability that people have to achieve a small data - understanding the big task of the foundation. As a result, artificial intelligence might be able to achieve truly intelligent