The problem of designing an AI to understand what it sees

@sonhouse said
@KellyJay
You are catching on. For instance, suppose at some future time, a computer is developed after centuries of development like we have seen to this date, so this computer is a thousand times smarter than any human. Suppose it is asked by humans to develop a bomb so powerful it can blow up an entire planet in a second.
Do you think said computer will go, sure, I alre ...[text shortened]... s one even more powerful, it can blow up the SUN.
Do you seriously think it would comply with such?

You are not getting my argument if you don't understand this one point, there is information within a computer, but it isn't the computer's it is ours. The computer is only a device, it does the work we assign it to, and it will only do what we tell it too. The computer will do what we say and do it quicker and quicker, but that is still nothing but the computer doing what we say the way we say it.

@kellyjay said
there is information within a computer, but it isn't the computer's it is ours.

The more advanced AIs already generate their own knowledge via learning and with that knowledge coming NOT from any human, let alone the human programmer, but from the AI learning process itself (I will give specific examples of that on request. The more advanced AI chess machines for example learn certain chess strategies never thought up by any human). Before if or when it shares that AI-learned knowledge with us, how is that knowledge "ours" and not "AI knowledge"?

@humy said
The more advanced AIs already generate their own knowledge via learning and with that knowledge coming NOT from any human, let alone the human programmer, but from the AI learning process itself (I will give specific examples of that on request. The more advanced AI chess machines for example learn certain chess strategies never thought up by any human). Before if or when it shares that AI-learned knowledge with us, how is that knowledge "ours" and not "AI knowledge"?

I promise you it is not due to the computer’s will, drive, hunger, and thirst for knowledge that is taking place! It is doing what we want by design and everything in it does with all of that will be at our design as well.

@kellyjay said
it is not due to the computer’s will, drive, hunger, and thirst for knowledge

So what? What has its absence of feelings got to do with the fact it has its own knowledge, which you denied?

@humy said
So what? What has its absence of feelings got to do with the fact it has its own knowledge, which you denied?

It has data stored in it; if it has data in it or not doesn't alter its knowledge, it doesn't know anything. It is a machine with components that has voltage applied so there is current directed through various parts and gives an output when told.

@kellyjay said
It has data stored in it; if it has data in it or not doesn't alter its knowledge,

Wrong; it can gain new knowledge from processing data. That new knowledge and, in a few cases I know of, even that data may not have been previously known by any human. And you sill haven't answered my previous question ( What has its absence of feelings got to do with the fact it has its own knowledge? ) so still have no idea what you where saying there.
You also seem to be implying it CAN have its own knowledge with your above words "...its knowledge" after clearly implying it cannot, so you clearly are being incoherent.
So now I really have NO IDEA what's the point you are trying to make in these latest posts.

@humy said
Wrong; it can gain new knowledge from processing data. That new knowledge and, in a few cases I know of, even that data may not have been previously known by any human. And you sill haven't answered my previous question ( What has its absence of feelings got to do with the fact it has its own knowledge? ) so still have no idea what you where saying there.
You also seem to be im ...[text shortened]... herent.
So now I really have NO IDEA what's the point you are trying to make in these latest posts.

It gains data. Computers do not know what data is there is no understanding within it for all of its processing power. It lacks all of the essential things to understand, from feelings to awareness of any kind, there is nothing in it that knows anything.

The very first computer I ever worked in was in the navy used those pain in the butt card feeders for communication. For home use, I don't recall the processor for games the data on cassette tapes (8086 maybe)! The technologies went to MMF, RLL hard drives. The first potent CPU I owned was had a 286 processor and on and on till now. Its all different mediums doing the same things at different speeds, you think these things also knew anything at all or were they just devices?

@kellyjay said
It gains data.

Got that.

Computers do not know what data is

If its an AI, in what sense "not know"? Define what you mean by "know".

It lacks all of the essential things to understand, from feelings ...

Apart from for understanding the feelings themselves, you CLEARLY don't need feelings to understand EVERY kind of something. For example, its just a trivial observation that I can understand some maths, or how a car works, without feelings involved.

...to awareness ...

Again, define what YOU mean by the word "awareness" here that is different from what I mean by the word "awareness" such that it makes no sense to say an AI cannot be 'aware'.... You see, there ALREADY exists AIs that have what I would call "awareness".

The very first computer I ever worked in was ...

STOP right there! I am talking about AIs here, NOT just your average dumb computer. Obviously, you Can validly argue that you average computer doesn't really 'know' anything, because it really doesn't. BUT that is NOT the claim of yours I dispute here but rather the implied claim that, EVEN if the computer is an extremely advanced AI, it cannot 'understand' anything! THAT is where you are wrong!

@humy said
Got that.

Computers do not know what data is

If its an AI, in what sense "not know"? Define what you mean by "know".

It lacks all of the essential things to understand, from feelings ...

Apart from for understanding the feelings themselves, you CLEARLY don't need feelings to understand EVERY kind of something. For example, its just a trivial ...[text shortened]... computer is an extremely advanced AI, it cannot 'understand' anything! THAT is where you are wrong!

They are one and the same, one is just an upgraded version of the others. Nothing new except how well they do the things the older ones did.

@humy said
Got that.

Computers do not know what data is

If its an AI, in what sense "not know"? Define what you mean by "know".

It lacks all of the essential things to understand, from feelings ...

Apart from for understanding the feelings themselves, you CLEARLY don't need feelings to understand EVERY kind of something. For example, its just a trivial ...[text shortened]... computer is an extremely advanced AI, it cannot 'understand' anything! THAT is where you are wrong!

Beyond the upgrades in devices and programming languages there is no change in what is taking place. An average dumb computer is still a computer without consciousness, our super computers are as well.

Some of the fast ones in the past were computers with several “dumb” CPU working together in them. What piece of hardware do you think we have made can take a computer and turn it into something aware of knowledge.

@sonhouse said
@KellyJay
You are catching on. For instance, suppose at some future time, a computer is developed after centuries of development like we have seen to this date, so this computer is a thousand times smarter than any human. Suppose it is asked by humans to develop a bomb so powerful it can blow up an entire planet in a second.
Do you think said computer will go, sure, I alre ...[text shortened]... s one even more powerful, it can blow up the SUN.
Do you seriously think it would comply with such?

Compliance and self-awareness are in conflict here. Why would a computer 1,000 times smarter than a human listen to anything we said? Do we obey squirrels?

@wildgrass said
Compliance and self-awareness are in conflict here. Why would a computer 1,000 times smarter than a human listen to anything we said? Do we obey squirrels?

You assume a computer “knows something” to be arrogant with. Processing power with the simplest pocket held calculators in many areas out preforms us, that intimidate you?

@kellyjay said
You assume a computer “knows something” to be arrogant with. Processing power with the simplest pocket held calculators in many areas out preforms us, that intimidate you?

Processing power is not intimidating. But machine learning and AI has already advanced beyond humans' ability to understand what they are doing. I think we should have safe guards in place to protect the internet from AI. Since AI tech is limited to only a few groups of people, they have a lot of power to direct AI at specific areas. It seems to me that independent regulatory oversight should be an important part of developing this technology, but that doesn't appear to be happening.

@wildgrass said
Processing power is not intimidating. But machine learning and AI has already advanced beyond humans' ability to understand what they are doing. I think we should have safe guards in place to protect the internet from AI. Since AI tech is limited to only a few groups of people, they have a lot of power to direct AI at specific areas. It seems to me that independent regulato ...[text shortened]... should be an important part of developing this technology, but that doesn't appear to be happening.

It is not learning in the same way we do; it does not know anything; it is merely doing what is programmed into it, only working towards the end goals of the programmers and operators.

Looking at all the possible chess moves from a position is not understanding chess; it is taking in a set of parameters and doing the calculations. If it is solving a problem great, but the computer itself does not know what a problem is. It will not know what chess is before or afterward.

@humy said
The more advanced AIs already generate their own knowledge via learning and with that knowledge coming NOT from any human, let alone the human programmer, but from the AI learning process itself (I will give specific examples of that on request. The more advanced AI chess machines for example learn certain chess strategies never thought up by any human). Before if or when it shares that AI-learned knowledge with us, how is that knowledge "ours" and not "AI knowledge"?

So one day all by themselves these AI on their own without a programmer telling them too just took it upon themselves to come up with better chess strategies never thought of before by any human?