I use to think so but noe I think unpredictable and cruel humans are scarier. You can debug/troubleshoot a machine but people's free will and complexity makes them dangerous.

If a robot sues me, I can sue its operator. If a person shoots me, they can claim an accident, self-defence,qualified immunity,etc... so robot operators have a good incentive to be right when they fire.

The scary part isn't the bots but who controls them.