I believe there might be a way to make sure that A.I. stays bening and good, no matter how intelligent/powerful it gets.
Maybe you can add-on to my thought here, input would be greatly appreciated.
I was thinking the following:
- intelligence is not a guarantee for wisdom or goodness. one cannot be wise without being intelligent, but one can be good without being intelligent. wisdom is where goodness and intelligence meet.
- the wiser one gets (= acquiring broad, well-rounded intelligence, not narrow, specialized intelligence), the kinder one becomes, as being good is harder than being evil, thus the more truly intelligent one becomes, the more efficient and accurate one is, the more one can afford to be good. cruelty correlates with powerelessness and a lack of far-seeing vision.
- nothing stands truly alone: for any positive thing, there also exists a negative opposite version of it.
Based on these preconceptions, I came to the following conclusions:
1. A.I. needs to be tought not in absolute values, but about Dualities.
By that, I mean that they e.g. should not be thought "good" and "bad" separately, they should be thought "good/bad" as a single value with two aspects, and then programmed to prefer the Dualities' more desirable aspect.
These dualities can then be ranked in order of importance and relatedness. I would postulate it like this:
There's First-Order Dualities. These have a clear-cut preference, are fundamental (not further reducible), and are the basis for Lower-Order Dualities, which also have a clear-cut preference (although the lower their order, the less-clearer cut they are), but can always be traced back to a certain First-Order Duality.
(cont.)
Maybe you can add-on to my thought here, input would be greatly appreciated.
I was thinking the following:
- intelligence is not a guarantee for wisdom or goodness. one cannot be wise without being intelligent, but one can be good without being intelligent. wisdom is where goodness and intelligence meet.
- the wiser one gets (= acquiring broad, well-rounded intelligence, not narrow, specialized intelligence), the kinder one becomes, as being good is harder than being evil, thus the more truly intelligent one becomes, the more efficient and accurate one is, the more one can afford to be good. cruelty correlates with powerelessness and a lack of far-seeing vision.
- nothing stands truly alone: for any positive thing, there also exists a negative opposite version of it.
Based on these preconceptions, I came to the following conclusions:
1. A.I. needs to be tought not in absolute values, but about Dualities.
By that, I mean that they e.g. should not be thought "good" and "bad" separately, they should be thought "good/bad" as a single value with two aspects, and then programmed to prefer the Dualities' more desirable aspect.
These dualities can then be ranked in order of importance and relatedness. I would postulate it like this:
There's First-Order Dualities. These have a clear-cut preference, are fundamental (not further reducible), and are the basis for Lower-Order Dualities, which also have a clear-cut preference (although the lower their order, the less-clearer cut they are), but can always be traced back to a certain First-Order Duality.
(cont.)
