A lot of concerns about future AI safety have been heard lately. A big one is that a supper intelligent (let’s say IQ>4000) machine intentionally kills people or unintentionally causes to wipe out the whole humankind.
To prevent the disaster from happening, one potential strategy/method is to democratize AI and let everyone get access to AI technology. Everyone can make or have a supper AI machine of capable to kill everyone in the planet. Therefore, a normal human being will not go down that road because he would be the victim too.
Not convinced: it is only a wishful thinking of a normal human being. Do not forget ISIS, Al Qaeda, Kim Jong Un, etc. Moreover, Americans “democratize” guns and lot of Americans die of gun shots each year.
If a supper AI machine would really be intelligent, is it manipulable by humans (good or bad)? Just think about it, can an amoeba manipulate us to kill someone?
On the other hand, humans do not need pandas but we also do not kill them, instead, we declare pandas are endangered species.
Similarly, supper AI machines do not need humans and much not so the earth but there is no obvious reasons that super AI would kill us. Probably, it would declare humans as endangered species and earth as reservation planet 🙂
Anyway, humans are notorious story believer (Yuval Harari). If someone/something (with IQ>4000) has a good story and would convince us he is God 2.0 and we humans would immediately worship the new God.
On the other hand, humans have a strong ability to learn things. Probably humans become also stronger/clever by learning from super AI and cope better with super AI?