what do you constitute as 'real' intelligence

[tyrantofthought]

In that meaning, what, to you, is the difference between intelligence and artificial intelligence? I personally find very little difference. Now I'm not saying artificial intelligence is as valuable as humans, but in truth dont bugs, rodents, and whatever else have intelligence? Is their intelligence more 'valuable' than a AI robots intelligence morally to you?

[Jakalwarrior]

artificial intelligence to me would be anything intelligent that is not alive. I also define alive as the ability to reproduce.
*Side note, what about a "fixed" human etc..? You are designed to reproduce, but are broken*


You make a brain in a computer you've made AI
You make a brain in a computer that can reproduce without your assistance you've made life! - Ability to change to better suit its enviornment and youve made an adaptable living organism.
If that creature also cares more about its own existence and reproduction than your well being, then youve made competition!

[EsoteriEccentri]

I also define alive as the ability to reproduce.


Do you mean aswell as all the living organisms? So you would include robots who can reproduce too?
Or JUST things that can reproduce. Infertile people aren't alive!! D=



I define "real" intelligence as thought. Ability to think about things, contemplate, think of new ideas, solve problems, examine concepts. Introspection would be a form of intelligence in the sense that animals cannot do it, (in my belief).
Other forms of intelligence unique to humans would be ability to philosophise, ability to imagine to the extent of writing stories, drawing art, ect.
I would also include ability to empathise.

I wouldn't include the ability to come up with completely new ways of doing things, solving problems, learning tricks, ect - as they are not unique to humans. But they are still forms of intelligence.

But intelligence comes in many different forms in humans (remember, no type is actually the most intelligent type), but also amongst other animals. Chimps excel us in some things. xD

To me, anything intelligent is anything alive though perhaps the level and type of intelligence can vary. Anything alive is anything that is mortal and can die. Unless, of course, we manage to become immortal (sure) and so I would change it to anything that can die OR anything with individual conscious thought. Which perhaps one day will include very advanced artificial intelligence. I guess it depends where you draw the line, at what point a machine is so complex you could really say it "thinks" and has a personality.

[Jakalwarrior]

Do you mean aswell as all the living organisms? So you would include robots who can reproduce too?
Or JUST things that can reproduce. Infertile people aren't alive!! D=


Why not? more advanced than a prion or virus. Even if it requires special materials to reproduce, then it is just a very gimpy, unadaptive life form :D
An infertile human being is just a broken reproducing organism. If all human beings couldnt reproduce though, then we wouldnt be alive ;) and if somone crafted a single human being from scratch without the ability to reproduce. What would be the difference between that and a highly advanced robot?

[unnu]

In that meaning, what, to you, is the difference between intelligence and artificial intelligence?

1. Computation - almost everything in nature computes, from quarks to galactic superclusters
2. Emotion - feedback mechanism which permits improvement in a given situation
3. Cognitive complexity - the ability to solve complex tasks, increasingly (high g, as measured in humans)
4. Discovery - a result of creativity; the method is, by definition, not known in advance (difficult to measure other than direct observation)

If the AI can autonomously function at all those levels, I would consider it equivalent to known forms of intelligence. But there is not yet AI with that ability, so it's hard to say what the difference is other than it's not yet 'intelligent'. :]

[sriv]

2. Emotion - feedback mechanism which permits improvement in a given situation

Why does one need emotion to improve?

[unnu]

Why does one need emotion to improve?

If you don't like the word emotion, it can be replaced with self-reference. How does one improve without being aware of something to improve?

[brad]

begs the question, what is the evolutionary biological role of emotions?

[sriv]

If you don't like the word emotion, it can be replaced with self-reference. How does one improve without being aware of something to improve?

An innate sense of idealism would accomplish the same purpose.

begs the question, what is the evolutionary biological role of emotions?

I would say that at first, they fused us together into a community. Back then the logic behind the idea of "unity is strength" was weak. Now we can establish that "unity is strength" empirically and logically so emotion is no longer necessary to use as a binder.
They made us either disappointed in the present which led to hard-working or too happy in the present which led to lethargy. They fueled us to achieve happiness and advance.

Other than that emotions evoke certain chemicals for certain tasks. Anger or fear give bursts of adrenaline and raise blood pressure. Evolutionarily, adrenaline is a huge advantage.

[tyrantofthought]

Well yay. I was hoping I wasnt the only one that thought intelligence could be 'made' by us in some way. I was thinking of it since I just finished watching all of a show that had some artificial intelligent robots that in the end were more or less alive by my standards.

[Monte314]

I have been doing work in artificial intelligence for many years. I have designed heuristic systems having embedded knowledge and the ability to improve their performance on assigned tasks... I don't regard them as intelligent, because I know how they work.

I once built a program for solving a certain kind of puzzle. I endowed it with the ability to enhance its performance over time as it played the game. Eventually, I was no longer able to beat it. I found that it had discovered a "winning strategy" for the game that I, ITS CREATOR, had never heard of. Once again, I don't regard it as intelligent, because I know how it works.

About three years ago, I built an application that plays checkers. Initially, it knows nothing but the rules, so it makes a lot of stupid moves and is easy to beat. But, as it gains experience, it modifies it's internal strategy, becoming more and more formidable. Helping it learn by playing against it was tedious, so I decided to allow it to play against itself. It can start from scratch, and play 5,000 games against itself in a few minutes, vastly increasing its skill. This, I thought, was cool. (I use it in my Artificial Intelligence class... the students love it, because they can see it getting smarter as the entire class tries to beat it!) I don't regard it as intelligent, because I know how it works.

I am now working on systems embedded in peoples' computers that "watch" them work, and learn to recognize their behavior patterns. These programs can often predict what the human user will do next. However, even though these applications can foresee intelligent behaviors, I don't regard them as intelligent, because I know how they work.

Did you see the common theme here? We will NEVER regard a mechanism whose inner workings are known to us as "intelligent", because in our hearts we believe that "intelligence is magic", and can't be present when we have a ready naturalistic explanation.

I was invited to give a colloquium on machine intelligence in Great Britain some years back... I gave my presentation, and there was one young man who hovered in the back of the room until everyone else had gone. I knew *exactly* what he wanted, because it's happened before.

He approached me as I was collecting my materials, and said,"Sir, I have a question."

"Yes?" (I knew the question already)

"Sir, how is it that intelligent machines REALLY operate? What principal allows them to exhibit human behaviors, learn, adjust themselves, and the like? How do they do this?"

I lowered my voice... "Do you REALLY want to know?" I almost whispered.

"Uh, yes. Yes! What is the secret?"

Here is the answer I gave him... and it is the correct answer. "If you take the most amazing 'intelligent' machine ever built, and crack it open, what you'll see inside is ZEROS and ONES. That's the secret."

That one exchange made the entire trip worthwhile! I am always amazed and amused that people actually believe that intelligent behavior arises from systems that don't have it designed in. The macro-evolutionist's "From goo to you by way of the zoo" is NOT an explanation. But, I digress...

Consider this example:

Suppose I have an amazing machine with a slot in front and a button on the side. I ask the machine a question: "What will I have for dinner tonight?" I push the button, and a silp of paper pops out that says, "Tonight, your wife is preparing pot roast. You will enjoy it with a fine Merlot."

Amazing, I think to myself. I call a dozen of my co-workers in. They pepper the machine with hundreds of questions, and each time, the machine produces a spot-on, cogent, grammatically flawless and completely responsive answer.

"Now THAT's AI!" they all say.

So, of course, we pop the thing open only to discover that it's nothing but a roll of paper with random sentences on it which have been correct, apprently, by RANDOM CHANCE.

We close the machine back up, and it once again produces an unbroken string of correct answers. But this time we all agree "That' NOT AI!!!"

The only difference between the two assessments of this amazing device? Our knowledge of its inner workings. And so it will always be. We believe in our hearts that "intelligence" is somehow mysterious and transcendent, and so WILL NEVER accept as intelligent anything that is otherwise.

[unnu]

"If you take the most amazing 'intelligent' machine ever built, and crack it open, what you'll see inside is ZEROS and ONES. That's the secret."

Leibniz as the first AI researcher? ;]

I am always amazed and amused that people actually believe that intelligent behavior arises from systems that don't have it designed in.

This raises the idea of environmental influence. In the case of humans, we are coupled with our environment, it does affect our intelligence. To give an extreme illustration, imagine the following. Person A is raised in a sensory-deprived environment, i.e., no contact with other humans, no language, no culture, no group dynamics to integrate. This person has the potential to be an intelligent human, but without proper sensory input they will not develop. What if we extend this to humanity as a whole. Let's say we had no culture transmission, for instance, everything that's been achieved in mathematics over the past 2500 years is suddenly lost - since in this scenario it's impossible to transmit knowledge from generation to generation. Exactly how 'intelligent' would we be? Sure, we still have the same brain, we have the genetically inherited potential, but to acquire the same knowledge within a single lifetime that's otherwise been accumulated for millennia is impossible. I think this is usually taken for granted in comparative analysis between human and machine intelligence; the AI cannot "stand on the shoulders of giants" as easily. The primary point is that something which exhibits intelligent behavior needs more than itself to be considered 'intelligent'. Unless we also consider potential intelligence such as the case of Person A. What will happen if we develop AI that can understand our accumulated knowledge as efficiently as the average human? We only have one example of intelligence, which is the life that exists on Earth. Perhaps we will consider AI intelligent only if it can understand our history, our collective knowledge system.

We believe in our hearts that "intelligence" is somehow mysterious and transcendent, and so WILL NEVER accept as intelligent anything that is otherwise.

This is probably true, although maybe intelligence really is a transcendent phenomena? We still only have one example of intelligence - human intelligence as the extreme form relative to other life. Human intelligence researchers can measure the g factor quite well, it's one of the most successful endeavors in psychology, unfortunately it's also presented as one of the most controversial due its social ramifications, despite the evidence. We can measure g, but we don't precisely know what g is! This adds difficulty to the discernment of 'real' intelligence and AI.

I don't think I was able to write down all the thoughts that your post inspired in me, but I agree with most of what you wrote! I just wanted to append some thoughts to hopefully stimulate more ideas.

[tyrantofthought]

I must say, thank you greatly for you replyMonte314. I must agree with pretty much all you say, but is there still this little possibility my imagination is trying to keep alive? In the said show i mentioned in the beginning throughout the two seasons the AI robots they had at first were really just as you said, but once the robots started to have 'malfunctions' they were relieved of duty and went on to lower jobs for the city or whatever and so their experiences changed even more. Before the end of the first season they completely disregarded a direct order, and by the end of the second season their programming were pretty much just guidelines and not something they were set in. Mind you, this had to do with a highly advanced society and the AI bots were apparently the latest, so its not like we are talking about simple-like AI.

So, do you think that in any possible way there could be some way that if an artificial intelligence were given the chance it could go above and beyond its programming and in a sense, reprogram itself.
Then again I think of the human mind as little different from a robotic mind made with organic material, or AI vise versa, and because of that my view on intelligence is that todays AI is merely a much much simpler form of what our brain can achieve.

[tp6626]

I think it is a near certainty that artificial intelligence will go above and beyond its programming and reprogram itself as you say. The reason I believe this is that there seems to be nothing prohibiting it, and the reason I think there is nothing prohibiting it is that it has demonstrably happened before. The genetic code of DNA is a digital code (I'm not talking metaphorically here either), and it has developed intelligence and complexity from simple beginnings over hundreds of millions of years. This has been through a series of stacked 'fortunate' 'accidents'. I'm sure with our help, building on past knowledge, we can get softwares making 'better' accidents, faster.

[athenian200]

In that meaning, what, to you, is the difference between intelligence and artificial intelligence? I personally find very little difference. Now I'm not saying artificial intelligence is as valuable as humans, but in truth dont bugs, rodents, and whatever else have intelligence? Is their intelligence more 'valuable' than a AI robots intelligence morally to you?

At this point, the difference is their ability. AI is limited and unable to solve certain problems that normal intelligence can. But I would say that AI is more valuable that rodent or bug intelligence, even if it's not as complete in some ways, because of its value and usefulness to us.

If the human intelligence and machine intelligence were equal, the difference would be that one was organic, and the other was inorganic. They would be of equal value, in my opinion.

[Aronnax]

The only difference between the two assessments of this amazing device? Our knowledge of its inner workings. And so it will always be. We believe in our hearts that "intelligence" is somehow mysterious and transcendent, and so WILL NEVER accept as intelligent anything that is otherwise.

So what happens when we develop a deep understanding of how clustered neurons process information? Or the extension of that, a way to engineer a computer composed of neurons?

Personally I'd be willing to concede the possibility of intelligence when a machine can pass the Turing test. At that point it's for the best to give a machine the benefit of the doubt.

[Monte314]

So what happens when we develop a deep understanding of how clustered neurons process information? Or the extension of that, a way to engineer a computer composed of neurons?

Personally I'd be willing to concede the possibility of intelligence when a machine can pass the Turing test. At that point it's for the best to give a machine the benefit of the doubt.
Excellent points.

We already have computational devices constructed of both natural and engineered neurons (e.g., Carver Mead at Caltech has been doing this since the 1990's).

Arronnax has stated well the fundamental point: I believe that other people are intelligent because of the behaviors they exhibit. This is the heart of the Turing Test: intelligence can only be assessed phenomenologically, the nature of the underlying mechanism notwithstanding.

[Beery Swine]

I define real intelligence as the ability to use a TiVo or DVR remote without looking at it.

[Homini Lupus]

I always tought that, given enough time to develop, we would have seen more and more complex AIs able to replicate all of human emotions and tought. It would not just be the developement of artificial intelligence but also of artificial motivations and blurred logic (not to get the AI stuck with NP problems). With time it seems likely to me that we will no more be able to tell the difference from the outside. The question to me has always been: are they going to have a soul (meaning with soul that self awareness wich is negated by solipsism, with no spiritual meaning)? It's quite unlikely that a clockwork calculator has one, but what about a computer based on an organic engineering (unlikely by now but possible in the future)? After all even our brain is a sort of machine, even if
its working is still obscure to us.