The solution is in developing effective Heuristics.
Once you can correctly assess whether people are better or worse off by certain actions you can define morality by what action has the greatest net positive effect on people and choose that course.
RE: Can AI be trained to assist with moral decision making?
You are viewing a single comment's thread from:
Can AI be trained to assist with moral decision making?
Some people got crucified for doing what they thought was best for everyone.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
That is clearly just a disagreement over what is the best outcome. That's why the heuristics have to be effective and provable - thus undisputable.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
No, not everyone bases morals on rationality. Consequentialists do but not everyone is a consequence based thinker. I used to think like you but I realize not everyone is going to accept this. In fact if you look at the data the majority of people on the earth don't think this way.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Not everyone does, but machines and AI will. Unless you are suggesting we build AI that thinks irrationally.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
No not at all. The AI is logical (not rational). The person who asked the question, trained the AI, tells the AI it's values, has to be rational for the AI to provide a rational answer.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit