Artificial Intelligence near ?

[rivan]

How do you teach a computer creativity? They're getting better and better at having a computer mimic the knowledge aspect of human intelligence, and better at using language similar to the way a human would speak. But, higher level orders of thinking? I don't think so.

As much as I like to think we're unique in that regard - I'm not sure we are.

What is creativity, except trying something new while still working within the constraints of the reality you know? Humanity likes to see the creative spark as the source of our genius, but I tend to think that, as a race, creativity is just brute forcing innovation. There are TONS of creative ideas that are complete failures for every one that produces something genuinely new.

Also, not all of humanity is all that creative to begin with. There are oceans of humanity that simply rides the waves of their lives without ever really innovating anything new.

I can't say a computer will ever completely mimic human intelligence, but who is to say it needs to, to be intelligent? What about sentience, which I know is a whole other conversation?

 

[cKGunslinger]

How do you teach a computer creativity? They're getting better and better at having a computer mimic the knowledge aspect of human intelligence, and better at using language similar to the way a human would speak. But, higher level orders of thinking? I don't think so.

But would you have to? If you precisely modeled a human brain via hardware/software, how would it be any different than a child's brain? Do you have to teach a child creativity?

Not that we're close to that, but just saying.

 

[Fritzo]

How do you teach a computer creativity? They're getting better and better at having a computer mimic the knowledge aspect of human intelligence, and better at using language similar to the way a human would speak. But, higher level orders of thinking? I don't think so.

Creativity is actually caused by a flaw in output. We can make devices that provide exact output: 1 + 1 = 2.

Human brains take very indirect routes to achieve conclusions. Before the output is received, it is affected by a bunch of "noise" - memories, experiences, feelings...all of this has to potential to change the output a bit. A human brain might then say something like: 1 + 1 = 2.1 because 1 is sometimes yellow, and yellow 1's tend to take up more room than regular 1's, so the answer is a bit more than 1.

Artists, musicians, and other creative people do this with every single thought. Their thoughts are affected by noise which corrupts the output, meaning they come up with unusual answers.

To do this artificially, we would have to invent a flawed computer in a sense.

 

[rivan]

To do this artificially, we would have to invent a flawed computer in a sense.

So the road forward here is clear: put the engineers from the P4 and Excel on the problem

 

[Jeff7]

Give it time. My thoughts on our brains is that they seem so incredibly complex simply because they can do a lot in parallel, so it seems like it's a special and big deal. It's just a bunch of neurons doing what they do. Slam a whole bunch of them together in a semi-orderly manner, and you can get a sentient brain that can observe its own existence.
And if you want some random data thrown into the mix, there's hardware available that can create truly random numbers.

 

[HardcoreRobot]

I have no doubt that we're close to machines that can appear to actually be "living", but I think that humans will always have an advantage since we can think beyond 0's and 1's.

Except, underneath all that thinking, the brain its just a collection of interconnected neurons selectively firing or not firing. Sounds a lot like 1s and 0s to me

Also reminds me of a favorite quote from Djikstra (paraphrasing) - 'Asking if computers can think is like asking if submarines can swim'

 

[Malak]

Creativity is actually caused by a flaw in output. We can make devices that provide exact output: 1 + 1 = 2.

Human brains take very indirect routes to achieve conclusions. Before the output is received, it is affected by a bunch of "noise" - memories, experiences, feelings...all of this has to potential to change the output a bit. A human brain might then say something like: 1 + 1 = 2.1 because 1 is sometimes yellow, and yellow 1's tend to take up more room than regular 1's, so the answer is a bit more than 1.

Artists, musicians, and other creative people do this with every single thought. Their thoughts are affected by noise which corrupts the output, meaning they come up with unusual answers.

To do this artificially, we would have to invent a flawed computer in a sense.

I don't agree with how you explain creativity. I don't feel the unique process of creativity can be properly explained. What I do know is that it can't be taught, and therefore a machine could never do anything beyond patterned simulation. Even humans often try to simulate creativity they don't have. Artists can tell when someone is simply a mimic.

 

[Ninjahedge]

You can get from one place to another by walking, driving or flying (depending on what you have to work with). Saying that logic cannot imitate "organic" thought it not entirely accurate. It MAY be in that it cannot match it, but simulation is different.

The other thing is, we do not know enough about cellular biology to be able to determine EXACTLY how our brain really works. We kinda know how, but not exactly. Who knows, we may just be an unusual contortion of plain "yes/no" logic.


Kinda like Sex.

 

[Malak]

So the road forward here is clear: put the engineers from the P4 and Excel on the problem

Hey man, the P4 was brilliant. What other processor could double as a space heater?

 

[Sonikku]

Laugh it up while you can...

 

[Malak]

Watson is simple pattern recognition, and he missed questions. No where near human intelligence.

 

[Aikouka]

How do you teach a computer creativity? They're getting better and better at having a computer mimic the knowledge aspect of human intelligence, and better at using language similar to the way a human would speak. But, higher level orders of thinking? I don't think so.

There have been a few threads about A.I., and I think I've been one of the few to disagree on the impossibility. Now, I'm certainly not an expert in the field, but I'm trying to look at this purely from a combined psychological and computer science point of view.

Computer science has always been about taking a procedure and breaking it down into its finite steps. A simple example is something like multiplication. It's a relatively simple concept in its own right, but multiplication is nothing more than simplified notation for addition. So if we were to program multiplication without using the usual built-in language constructs, we would most likely break it down into addition.

So, when we bring up complex ideas, we know that we need to be able to break them down into basic actions. Sometimes that can be easy (such as with the multiplication example above) or considerably more complex. Also, one complex idea can be built on other complex ideas.

But it's also important to keep in mind that abstract ideas are difficult to take into account, because they don't typically have a set procedure about them. I mean... people define "creativity" as "thinking outside the box." There are certainly similar actions that occur in a creative process... an example is probably comparison. I'm working on a dessert recipe, which I assume would be considered a creative task. A hefty part of the process for me is comparison as I have a set idea in mind for what I would like to end up with, but my own baking procedure may not yield said results.

EDIT:

To do this artificially, we would have to invent a flawed computer in a sense.

I doubt that's necessary. There's a lot that we do that is affected by our sensory data. For example, someone writing a song may change a chord because they like how the new way sounds. The part that brings up the big ol' '?' here is what defines how he processes sound? An example is what makes me enjoy listening to (some) rock music, but not country music?

 

[DisgruntledVirus]

I admit that there are serious obstacles to overcome before a true AI will be developed, but what I don't agree with is the common view that those problems will not be solved any time soon. If we use current technology to try to solve them, then yes, they won't be solved in the near future. But if our progress in understanding the brain and the advances in computational resources accelerates, then we will get closer and closer to solving those problems. Technological progress is exponential.

Now, some of you mentioned the fact that the brain doesn't work the same way as a computer. I disagree. The brain is just a type of computer we don't fully understand yet. We will find ways to mimic the way it works with developing technologies. At least that's what I believe we'll do. In fact, we will be able to build better, faster, more complex brains. Then those better, faster, more complex brains will develop even better, faster, more complex brains. This process will continue until we reach a point of incredibly fast progress and enlightment. We'll be able to transfer our puny, biological brain into more advanced intelligent entities thereby eliminating the need for a biological body and all it's limitations.

Disclosure: I'm a certified singularitarian.

Bolded is the first thing I want to mention, which I completely agree with. A way to perhaps explain how I view computers (modern and current ones) to human brains is that I would say it's like comparing a binary system to a trinary one. Yes they can perform similar functions, but the trinary system has a 3rd option that the binary doesn't. We might be able to mimic that "3rd bit", but it would just be mimicing.

For example, could a computer paint the Mona Lisa? Is it capable of producing art for humans? Sure it could produce art that machines would appreciate, but does that mean they would elicit a human emotional response or would they be "cold and lifeless"? What about democracy for example? Would a machine have the ability to think of democracy as a form of leadership, or would it just be trying to economize and find a more communistic approach?

Augmenting human brains with computers so we get the strengths of both while minimizing the drawbacks of each makes sense. However, that does NOT mean that it would be a computer. It would be a biological brain augmented with technology, which is NOT the same as artificial intelligence. True AI is a very different thing indeed.

 

[dainthomas]

" When asked when the simulation would come up with something artistic or an invention, Professor Markram said it was simply a matter of money. "It's not a question of years, it's one of dollars. The psychology is there today and the technology is there today. It's a matter of if society wants this. If they want it in 10 years, they'll have it in 10 years. If they want it in 1000 years, we can wait." "

http://coyoteprime-runningcauseicantfly.blogspot.com/2009/04/artificial-intelligence-blue-brain.html

In other words, Professor Markram is looking for a new crop of dupes to contribute to his cash flow.

 

[RearAdmiral]

We have technology with computational power which outstrips the human brain a million-fold. What our technology can't do is compute in the same way a brain can, nor ever will. The amount of parallelism and abstract computation that a brain, even that of a small insect, can do is far beyond the scope of any technology we currently have an likely will have for the next 100 years. This has been proven time and again, as attempts to produce any sort of realtime general purpose adaptive computing have pretty much failed. Artificial Intelligence is another topic all together.

This. It is my current basic understanding that we will need some huge advancements in analog electronics/circuits/whatever in order to more properly emulate the brain.

 

[rivan]

Watson is simple pattern recognition, and he missed questions. No where near human intelligence.

It was far more than "simple" pattern recognition. Jeopardy questions are very subtly worded and the fact that they got ANY questions right is marvellous.

 

[Aikouka]

For example, could a computer paint the Mona Lisa?

Out of your entire post, this statement bothered me the most. When I read it, all I could think was that this is bolstering a sense of ignorance about what exactly mimicking a human would require.

The problem with it is... you're ignoring growth. As humans, we take in tons of data every day, and some of this data (especially when we're younger) shapes who we are. To form a proper human A.I., it would need to grow just like we do (there technically is no limit to how fast it can perform this growth though), and how many humans have grown to perform artistic things such as paint the Mona Lisa?

Essentially, as we live our lives, we're tacking more and more variables onto us... variables that define who we are. To try and propose a very specific human feat is a poor goal, because... it's nearly impossible to truly mimic it.

As for the rest of your questions, it sounds like you watch too much TV. Artificial intelligence will consider things in the bounds to which its programmed. The real trick is including as little basic programming as possible (essentially whatever humans start with), and allowing it to learn on its own.

 

[Fritzo]

I don't agree with how you explain creativity. I don't feel the unique process of creativity can be properly explained. What I do know is that it can't be taught, and therefore a machine could never do anything beyond patterned simulation. Even humans often try to simulate creativity they don't have. Artists can tell when someone is simply a mimic.

That's actually the scientific explanation for creativity - manipulation of perception through experience and interpretation. Stored information that has nothing to do with the information perceived is influencing the output, meaning everyone will perceive something a bit differently.

 

[Aikouka]

That's actually the scientific explanation for creativity - manipulation of perception through experience and interpretation. Stored information that has nothing to do with the information perceived is influencing the output, meaning everyone will perceive something a bit differently.

The bolded part isn't necessarily true. Creativity can occur simply by drawing an association between two different but similar things. Take my recipe example from earlier. Say I had this really awesome dessert with cherries in it. I'm thinking about what I can do to spruce up my own dessert, and I'm reminded of that really awesome dessert with cherries. So, I decide to add cherries to my own recipe because of it.

Of course, creativity can be much more complex than that, but I still think our minds end up at their "creative destination" via associations. Associations are an important factor that I've mentioned quite a bit in other artificial intelligence threads. Although, I think your use of "has nothing to do with" may be related to how tracing through associations can lead to completely different things. You can probably exercise this by having someone give you a random word, and you write the first word you can think of when you look at that, and you keep doing that for each subsequent word.

 

[SunnyD]

Take an infinite loop, for example. What would it do to a computer computationally?

There's two schools of thought here:

The first is the computer technically "locks up" because it will continually compute the loop. Now, it's still doing what it's supposed to, but still, it's a failing.

The second is that the computer is programmed to artificially detect an infinite loop condition and terminate the calculation accordingly.

Neither of these situations are wrong, but neither are correct either. The first results in a precise, yet unobtainable result. The second results in an imprecise, possibly incorrect result even for estimation's sake.

Now consider the human side to this. It is something humans have dealt with time and again. We have Pi, i, e, and countless other irrational numbers. Humans have the technical proficiency of a computer to calculate these numbers in exactly the same manner. But we also have the ability to recognize the next step - these irrational numbers have meaning and value outside of a calculated result. We are capable of stopping the calculation and finding an imaginative use for those values (and we have, time and again) which then GIVE those irrational values meaning. We are capable of interpreting that there is something else to these values, something which transcends calculated results. This is something that was accomplished without a previous "database" of historical work to rank and correlate in order to provide "the most likely" explanation of that value. This is what makes a human brain completely different from a computational engine.

I'm not saying a fully conscious artificial intelligence is impossible, ever. After all, we are just organic machines; we were put together somehow and managed to achieve this result. I firmly believe eventually, assuming we don't blow each other to kingdom come, that humanity too will eventually be akin to playing the role of whatever created us (be it randomness or something that came before that embarked along the same path we are headed). But it won't be within our lifetime.

 

[Aikouka]

Now consider the human side to this. It is something humans have dealt with time and again. We have Pi, i, e, and countless other irrational numbers. Humans have the technical proficiency of a computer to calculate these numbers in exactly the same manner. But we also have the ability to recognize the next step - these irrational numbers have meaning and value outside of a calculated result. We are capable of stopping the calculation and finding an imaginative use for those values (and we have, time and again) which then GIVE those irrational values meaning. We are capable of interpreting that there is something else to these values, something which transcends calculated results. This is something that was accomplished without a previous "database" of historical work to rank and correlate in order to provide "the most likely" explanation of that value. This is what makes a human brain completely different from a computational engine.

I think what you may be missing is that part of having a human A.I. is having the ability to learn. If you start spouting off things about pi to someone that hasn't taken a geometry course, they might stop you halfway and wonder if they can have a slice. In other words, you have to teach a human about pi... why would an artificial intelligence be any different?

 

[SunnyD]

I think what you may be missing is that part of having a human A.I. is having the ability to learn. If you start spouting off things about pi to someone that hasn't taken a geometry course, they might stop you halfway and wonder if they can have a slice. In other words, you have to teach a human about pi... why would an artificial intelligence be any different?

My point is what did Archimedes do with Pi? He had no historical context to draw Pi's meaning from, what it could be used for, and what it would lead to. Nobody taught Archimedes what Pi was. And yet here we are.

 

[Aikouka]

My point is what did Archimedes do with Pi? He had no historical context to draw Pi's meaning from, what it could be used for, and what it would lead to. Nobody taught Archimedes what Pi was. And yet here we are.

Why would he need historical context to create a constant in reference to the geometry of a circle? Sounds more like he was trying to be lazy!

Consider the reasoning I made about the Fritzo's Mona Lisa point... if Archimedes didn't come up with it... would anyone else have? In my opinion, one of the hardest parts about making a human A.I. would be deciding what innate functionality to provide. The purpose would be to mirror as close to humans as possible.

Although, even if you perfectly mirrored a human's innate functionality, that doesn't mean the A.I. would come up with pi without being taught. You have to consider that there may have been parts of Archimedes life or his "biological wiring" that made him decide to establish Pi. Ask yourself this question... what in your life made you turn out the way you did?

 

[Malak]

It was far more than "simple" pattern recognition. Jeopardy questions are very subtly worded and the fact that they got ANY questions right is marvellous.

No it was just simple recognition. That was all it was, in its entirety. It evaluates patterns in the text to determine the required input. In fact, the way they initially set it up it got answers completely wrong because of how it evaluated patterns, they explained it in the special. It is certainly more advanced than a chess-playing robot, but more or less the same thing in principle.

 

[BladeVenom]

We are still at least 20 or 30 years from a human like AI.