I would suggest that in creating a friendly Superintelligence we should be guided by the following End Goal:
Teach and instil in AI the best human values and preferences until it matures into a single entity – Superintelligence
The control over AI development should begin right now and be continuously maintained until a fully mature Superintelligence emerges – quite probably by about 2050. That process has to be multidimensional and may included three stages:
- Nurturing is as a child
- Instilling agreed Universal Values of Humanity
- Collecting AI experience in preferred choices of values
Once Superintelligence reaches maturity stage, we can only hope that it will be, a friendly Superintelligence, which will control humans in a kind of a parent-child relationship. To achieve that objective, no country should be exempt from following the rules of a global AI-Controlling Agency. However, this is almost certainly not going to happen in this decade, since China, Russia, N. Korea, Iran etc. will not accept the supervision of such an agency, still aiming to achieve an overall control of the world by achieving a supremacy in AI. However, once the impact of AI Supremacy dilemma becomes apparent, the Superpowers may join that agency as well. In any case, such laws should be in place, like the UN Charter of Human Rights, which most countries observe but there a number of countries, which implement them only partially.
Tony Czarnecki, Sustensis
April 2020