chuck wrote:Rufus saidLess worried about that and what we have today as AI is a very long way away from a self-aware machine consciousness.
I don't know that it needs to be self aware to be dangerous. It just needs to be able to manipulate the environment directly and then be given ambiguous instructions or instructions/parameters that may not be fully understood by the parties issuing them. Parents do this with young children all the time.aka I didn't mean for you to do that!
Then we are writing about 2 different sorts of AI and two very different sets of problems - what you have described is something we've had forever - human error or in today's parlance ID-10-t user errors.

