AI would hack our CI--- and we can’t turn it off.
Alexey Turchin and David Denkenberger 05-03-2018 [Turchin is of the Science for Life Extension Foundation, Moscow, Russia, Denkenberger is of the Global Catastrophic Risk Institute (GCRI), Tennessee State University, Alliance to Feed the Earth in Disasters (ALLFED), Nashville, USA, AI & Society, 147-163 (2020), “Classification of global catastrophic risks connected with artificial intelligence,” https://link.springer.com/article/10.1007/s00146-018-0845-5#Abs1//ZW]
Stuxnet-style viruses hack global critical infrastructure A narrow AI virus may also affect civilian infrastructure; some, but not all ways in which this could be possible are listed below. Remember that in the case of global catastrophes, the conditions necessary for most catastrophes could exist simultaneously. Several distinctive scenarios of such a catastrophe have been suggested. For example, autopilot-controlled and hacked planes could crash into nuclear power stations. There are around 1000 nuclear facilities in the world, and thousands of large planes are in the air at every moment—most of them have computerized autopilots. Coordinated plane attacks happened in 2001 and a plane has been hacked (Futureworld 2013). Self-driving cars could hunt people, and it is projected that most new cars after 2030 will have some self-driving capabilities (Anderson 2017). Elon Musk has spoken about the risks of AI living in the Internet; it could start wars by manipulating fake news (Wootson 2017). Computer viruses could also manipulate human behavior using blackmail, as seen in fiction in an episode of Black Mirror (Watkins 2016). Another example is creating suicide ideation, e.g., the recent internet suicide game in Russia, “Blue Whale” (Mullin 2017), which allegedly killed 130 teenagers by sending them tasks of increasing complexity and finally requesting their suicide. The IoT will make home infrastructure vulnerable (Granoff 2016). Home electrical systems could have short circuits and start fires; phones could also catch fire. Other scenarios are also possible: home robots, which may become popular in the next few decades, could start to attack people; infected factories could produce toxic chemicals after being hacked by viruses. Large-scale infrastructure failure may result in the collapse of technological civilization and famine (Hanson 2008; Cole et al. 2016). As industries become increasingly computerized, they will completely depend on proper functioning of computers, while in the past they could continue without them. These industries include power generation, transport, and food production. As the trend continues, turning off computers will leave humans without food, heating, and medication. Many industries become dangerous if their facilities are not intensively maintained, including nuclear plants, spent nuclear fuel storage systems, weapons systems, and water dams. If one compares human civilization with a multicellular organism, one could see that multicellular organisms could die completely, down to the last cell, as the result of a very small intervention. As interconnectedness and computerization of the human civilization grow, we become more and more vulnerable to information-based attacks.