Remember, once in the information noise slipped the news about the Chinese broadcaster, which controls the way a neural network? This TV host in the format capable of nonstop news program 24 hours a day, 7 days a week.
Actually, nothing surprising. We have long been accustomed, that the world of technology is developing rapidly.
However, the presenter plays the news that it somehow serves the people. Yet the neural network is trained to emulate the normal host, but to highlight interesting news and process the visuals they still can’t…
But not so simple.
For example, the American company OpenAI developed algorithm language modeling GPT-2. The possibility of artificial intelligence using this algorithm is small: the output is connected short stories, or invented by the neural network, a small story on a given subject.
The algorithm OpenAI refuse to publish, because its use can adversely affect the quality of the content that will be published on the Internet. The drop in the quality of content – this is not the worst scenario.
Imagine a neural network – rewriter. Having trained in the content of real people, artificial intelligence will be able to generate a huge number of related in meaning, but almost unique texts. Because the neural network involved in the context, unlike sinonimayzer.
But most interesting is that the neural network trained by the algorithm can be used in the role of copywriters. Artificial intelligence can easily generate gigabytes of content just starting from a given starting theme.
And this, incidentally, is a direct opportunity to use it to generate fake news to create a factually inaccurate opinion in social networks, generate a myriad of different content comments.
In General, when the source GPT-2 will get to acess the Internet will change dramatically.
And it’s ever going to happen.