The problem of designing an AI to understand what it sees

@KellyJay
If said AI was to be somehow transferred to a human body and mind, how would you tell it was man made?

@sonhouse said
@KellyJay
If said AI was to be somehow transferred to a human body and mind, how would you tell it was man made?

The fact it had not morals would be a clue. If we are going to leap into make believe I'd say you can just say whatever you want now we have left reality.

@KellyJay
Why do you assume it would not have morals?

@sonhouse said
@KellyJay
Why do you assume it would not have morals?

How would it, can you program morals?

@kellyjay said
How would it, can you program morals?

If morals are a set of rules then it's quite easy. A better question is whether an AI can be produced which has affect, and whether our emotional artificial intelligence would have empathy for the entities that created it.

@deepthought said
If morals are a set of rules then it's quite easy. A better question is whether an AI can be produced which has affect, and whether our emotional artificial intelligence would have empathy for the entities that created it.

Really tell how you would program good from evil?

@kellyjay said
Really tell how you would program good from evil?

Just tell the AI (assuming it is advanced to understand such concepts, and it is just a matter of when, not if, one will be made that can) evil is things like murder, deliberately causing pain and fear and persecution and oppressed for being black or for not believing in a particular religion etc.
Good is preventing deaths and pain etc and making the lives of people more pleasant and feel more meaningful by various means, such as by giving them a good education and make sure they don't go hungry and allowed freedom of thought rather than told they must believe there is a god or else, etc etc.
What's the problem here? I don't see one.

@humy said
Just tell the AI (assuming it is advanced to understand such concepts, and it is just a matter of when, not if, one will be made that can) evil is things like murder, deliberately causing pain and fear and persecution and oppressed for being black or for not believing in a particular religion etc.
Good is preventing deaths and pain etc and making the lives of people more pleasa ...[text shortened]... told they must believe there is a god or else, etc etc.
What's the problem here? I don't see one.

That is the point the concepts, how would you build those?

@kellyjay said
Really tell how you would program good from evil?

It just follows the rules one's given it. The problem is that we're talking about these things as if they're algorithms in the traditional sense. The type of AI humy is talking about is something more like a brain simulation. A brain simulation would make these judgements in the same way a brain does.

@deepthought said
It just follows the rules one's given it. The problem is that we're talking about these things as if they're algorithms in the traditional sense. The type of AI humy is talking about is something more like a brain simulation. A brain simulation would make these judgements in the same way a brain does.

Except I don't think anyone has come up with how morals entered into our thinking by materialist processes! Especially if all the universe has going for it is stronger of the fittest method of species moving forward. Considering the weaker, which is, the younger and elderly doesn't suit this type of driving force in the universe.

I agree a lawgiver is required, but we cannot give ourselves a standard method of morals, so who could write the software for an android or robot? It is a can of worms if this topic is continued.

@kellyjay said
That is the point the concepts, how would you build those?

If you want to know then study AI.
Many AIs, such as one AI I created, already have concepts; -don't see what's the problem you think exists here.
Concepts can consist of logical relations and/or logical proposition (if you don't what those are, start here;
https://en.wikipedia.org/wiki/Logical_relations
https://www.britannica.com/topic/fact-value-distinction

also see
https://en.wikipedia.org/wiki/Knowledge-based_systems
-which is the thing I am an expert on.
)

@deepthought said
If morals are a set of rules then it's quite easy. A better question is whether an AI can be produced which has affect, and whether our emotional artificial intelligence would have empathy for the entities that created it.

Morality is more than a set of rules it’s the ability to justify judge between various beliefs, opinions, notions and coming up with the “proper response”, when even two right choices are at stake, and at odds with one another.

@kellyjay said
Morality is more than a set of rules it’s the ability to justify judge between various beliefs, opinions, notions and coming up with the “proper response”, when even two right choices are at stake, and at odds with one another.

Note the conditional at the start of my last but one post. Possibly I tend to conceptualize this differently. I see ethics as being the ability you're referring to. Morality is a codification of this and tends to appear as a set of rules, rights are an attempt at a codification of how one can expect to be treated by others at a minimal level.

@deepthought said
Note the conditional at the start of my last but one post. Possibly I tend to conceptualize this differently. I see ethics as being the ability you're referring to. Morality is a codification of this and tends to appear as a set of rules, rights are an attempt at a codification of how one can expect to be treated by others at a minimal level.

Ravi Zacharias I believe was the one who was talking about loving one’s neighbors and discussing the differences between people. He brought up some cultures believe in doing good as they want done to them, and then some eat them, asking if there was a preference? With these extreme examples and all points in between, a common framework of good and evil is difficult without a standard shared by all.

@humy said
If you want to know then study AI.
Many AIs, such as one AI I created, already have concepts; -don't see what's the problem you think exists here.
Concepts can consist of logical relations and/or logical proposition (if you don't what those are, start here;
https://en.wikipedia.org/wiki/Logical_relations
https://www.britannica.com/topic/fact-value-distinction

also see
https://en.wikipedia.org/wiki/Knowledge-based_systems
-which is the thing I am an expert on.
)

I have no doubt they do have concepts that there is plenty of things you can program into the systems; you can set up how it handles and relates one set of information with another. Your logic could be flawless, and still, there will be no understanding within the AI, it will remain subject to just what put into it, and it will do what it is programmed to do. There will be nothing about it that goes beyond its human design, no life of its own, no grappling with thoughts about itself, only those things by design and input will be there.