Читать Scientific research certifies – 4 - Андрей Тихомиров

The developing artificial intelligence (AI) can become so superintelligent and powerful that it will be able to lead to massive social upheavals and even to the destruction of humanity, The Week reports. According to the publication, more than 350 researchers and engineers have recently issued a warning that artificial intelligence poses a danger comparable to the risks of pandemics and nuclear war. In a survey of experts conducted in 2022, the average probability that artificial intelligence will cause extinction or a serious decline in humanity was 1 in 10. Jeffrey Hinton, who is often called the "godfather" of artificial intelligence, does not consider such concerns to be fantastic. Hinton used to think that the danger from artificial intelligence would appear in at least 30 years. However, now artificial intelligence, in his opinion, is developing into a superintelligence so rapidly that in five years it can become smarter than humans. ChatGPT based on artificial intelligence and Bing chatbot can already pass exams for obtaining law and medical licenses, and score 99% in IQ tests – that is, reach the level of human genius. Hinton and other experts fear the moment when artificial intelligence will be able to surpass humans in almost any task. Some artificial intelligence experts compare this possibility to the sudden arrival of a superior alien race on our planet. So, another pioneer in the field of artificial intelligence research, computer scientist Stuart Russell said: "You have no idea what they're going to do when they get here, other than they're going to take over the world." One of the apocalyptic scenarios is that attackers can use the capabilities of artificial intelligence to create new biological weapons that are more deadly than natural pandemics. Another danger is that as artificial intelligence is integrated into global systems, terrorists and other attackers can use it to disable vital infrastructure and financial markets. In addition, as indicated in the article, artificial intelligence may want to become completely independent and for this to completely get rid of human control. Therefore, artificial intelligence can trick the leaders of nuclear states into believing that the enemy has launched missiles, thus forcing them to launch their own missiles. Scientists believe that the uncontrolled rapid development of artificial intelligence, neural networks, and chatbots is dangerous for humans.