The development of a neural network can become more dangerous for humanity than nuclear weapons. This opinion shared Israeli historian and writer, author of “Sapiens: A Brief History of Humankind” Yuval Noah Harari in his article for the Economist.
The expert noted that insufficient control over the development of neural networks can play a cruel joke on humanity. The program can develop without the participation of people, unlike the development of nuclear weapons.
So, artificial intelligence can learn to create a better version of itself. AI has already learned how to compose sentences and form text from them. And language, as Harari noted, is what creates the culture of mankind.
In his opinion, the program may begin to misinform people with “political content and fake ironic news.” The author of the article noted that humanity should use such a powerful resource for good, for example, to solve environmental problems or develop a cure for cancer.
In mid-April, Chinese scientists gave AI control over the Qimingxing 1 space satellite for a day. The program chose in its opinion the most interesting points of the Earth and instructed the apparatus to study them. They coincided with the interests of the Chinese military. According to scientists, not a single person interfered with the experiment.