Thank you for your reply @khussan. Once strong AI is developed it is definitely going to be self sufficient, and you are right, that may cause plenty of problems (as well as opportunities). As I have written in the article, we need to ensure that stron AI will have human values implemented (morality etc.)
You are viewing a single comment's thread from:
I think morality cam be implemented by setting rules like action(kill)= illegal action(feed the cat) = legal
Posted using Partiko Android
More or less you are right @khussan, but that definitely wouldn't be so easy. AI learns itself through machine learning, which means it constantly modifies itself in order to achieve "perfection". Morality would have to be implemented in such way, that AI couldn't modify it.
Yes it has to be implemented in a hardcoded section i think can be achieved as a read only system/read only permission
Posted using Partiko Android