A superintelligence is by definition beyond what we can understand, therefore we will never know if it can be safe, it is better to build Narrow AI systems than an AGI Superintelligence