The "AI" industry is now claiming that its products could lead to the extinction of the human race and that we are all doomed. If this comes to pass, it won't be because "AI " programs have become psychopathic; it will be because human beings will have willed it and organised it.
"AI" is just a a highly organised algorithmic program which performs mathematical and decision table calculations at such fast speeds that it is beyond human comprehension. They are not self-motivated even if they are autonomous. They have no feelings, no moral compass, they are neither beneficent or maleficent. They do not think in human terms.
At the moment they are just tools which can either be used for the good of mankind or the bad of mankind.
The development of nuclear weapons can be used as an analogy. A nuclear bomb is created by the use of nuclear fissile and nuclear fusion materiel. Once the chain reaction of nuclear fusion is started, it happens very quickly and it is very difficult to stop.. Unimaginable amounts of energy can be be released in a split second to produce an explosion that is the equivalent of thousands or even millions tonnes of TNT being detonated. These weapons have been used already and it is within the bounds of possibility that they will be used again in a war that could destroy the whole of humanity. If humans had not invented such weapons they could not be used. It is not the actual bombs themselves which are the true danger, but it is the humans who have indulged in arms races, to produce such dangerous weapons. Nuclear bombs were invented by fools who could not appreciate the consequences of their actions.
We are told by politicians and nuclear entrepreneurs that nuclear energy can be used for the good of mankind to generate controllable energy, and this may be so. However, the use of nuclear power should have been limited to producing controllable nuclear power not uncontrollable bombs.
The development of "AI" or Machine Learning is at a completely higher level of utility to nuclear power. "AI" can inveigle itself into nearly all the aspects of human life. Unfortunately, its greatest and most serious use is seen in the military arena. "AI" is used to perform war games, and control battlefield strategy and tactics. It can be used to spy on adversaries, decrypt their messages and generate propaganda and false news. Robotic programs can be used to "identify" combatants and terrorists from their 'phone conversations and electronic communications, and then send attack drones to their locations and to kill them automatically. This can now be done by "AI" acting almost autonomously, but under the the general direction of its political and military masters. Of course, these autonomous programs can and do make mistakes and often kill innocent civilians and children who happen to get in the way of the explosive power. Politicians seem happy to allow unrestrained "AI" systems to be developed which can attack an adversary from a military or financial point of view. There is thus a race to gain the upper hand. "AI" systems will need to be able to both attack their adversaries and defend defend themselves against both physical and electronic countermeasures.
We have been lead to believe that "AI" is harmless but just like nuclear power in the wrong hands it could be extremely dangerous. "AI" software and it s agents should have been designed not to harm humans or their activities, and they should have been designed to protest human life and protect human dignity. Unfortunately, the malign genie is out of the lamp with regard to military applications.
We have already seen the ill effects of poorly designed and tested social media programs which encourage users to harm themselves and others and promote terrorist propaganda. Social media programs can be used by criminals to commit fraud. There are programs which can generate false images. "AI" programs can be used to write essays so that students can cheat. There are many useful aspects of social media but in my opinion these are more than counterbalanced by the malign aspects of the programs. We need to be very careful of the more malign aspects of the technology.
Perhaps, the most disturbing features of "AI" are their ability to attack "AI" operations of competitor states. "AI" could be used to decrypt financial messages thus rendering an enemy state's financial systems broken and useless to the detriment of citizens whose money and bank accounts could disappear. Robotic warfare could be controlled by "AI" to commit genocide and crimes against humanity. A programming mistake could cause a tit for tat robotic war which would be difficult to stop as events could take place too fast for human responses to stop the events.
