Artificial intelligence and chatbots

[erdos_suitcase]

Hi guys

Due to some research that i am undertaking i have started to look into AI and chatbots.

This has included looking at the Artificial intelligence markup language (aiml) and its interpreters.

Yes, in the case of the research that i am undertaking this should be sufficient , but i can remember creating this type of software on my zx81 back in the eighties....it seems that the software is very limited and could hardly be deemed artificial intelligence.

i can understand that general conversation is a matter of statement ..response..statement ....response , all of which normally revolve around a particular topic..but could we ever deem a machine to be truley Intelligent and apart from the turing test how would we measure the intelligence of a machine?

so the debate is can a machine be intelligent

and

How do we measure this intelligence?

 

[darkstang]

Currently the best artificial intelligence engine I've created is one that plays you in tic tac toe. And really... it's entire knowledge comes from a simple mathematical algorithm. There's only one way to beat it, that I have found. So from my experience... yes, we can create AI. But really it can only do things that are pre-programed based on whatever logic we give it.

How do we measure it? I'd say the best way would be to estimate the capacity it has against the average human, in whatever area you're looking for. An example of what I mean is... My tic tac toe program is unbeatable if it goes first, will tie itself every time, and can only be beaten one way. This is roughly the same as the average human player so I'd say it's .95 to 1 human.

 

[cointoss]

I remember a few years back when I was constantly high, I went to Google and mistyped a phrase I was searching for. It rearranged and corrected my spelling and asked me if that's really what I meant to search for. I was rather freaked out. I realize that at this point AI is still at a basic point (but it still beats my ass on most games...), but I fear if Humans continue to create machines and indeed intelligence to perform common tasks (and tasks we are too lazy to perform ourselves), we will one day be replaced...

 

[legendphil]

hao!

intelligent machines?. hmm.. some matrix stuff, but less romantic.

I still believe that machines cannot be as intelligent as men, theoretically we can do multiple processes simultaneously(computers single task but are very fast), think for ourselves, imagine, etc... things that machine can't do (although in time machines may <- shouldn't be allowed). We're still powerful, and over them, we should always be.

as to how to measure intelligence... I don't know... if machines would give us peace (without killing the human race), I'd consider them intelligent. PEACE!

 

[redlack]

to the Last post just above mine... It is human nature to at times be ilogical, impulsive, Arrogant, Conflicting, Unstable, and rude. For there to be peace we would have to either all to one religion (Take a pick there are to many good ones and so many bad/fake ones) wich would never happen, All place ourselves in solitary confinement with only occasional human contact for obvious social and :naughtyther reasons, Or be forced into this (Gilded cage is still a cage), in any of those instances the Entire Human race would be at war with eachother and the AI's, 2 logical conclusions result from this. Either Computer AI's wipe us out, or we beat them back down. In the first we all die, in the second we all get so scattered that we become nomadic again and are back to the stone age socially, keep in mind there would probably still be technology around.


Now about the thread in general... For any computer to be sufficiantly deamed intelligent to the Human degree It would need to be able to make Statements, questions, and theories. It would have to not only respond to outside stimulus, but be able to generate its own stimuli to influence the outside environment. It would have to be able to think, reason, and attempt to better itself by the example of others adapting to its environment in ways not thought of by its designer... and about a thousand other little things that i can't think of right now.

 

[taekwondokid42]

One thing I would like to point out here.

The human race has always been the most intelligent things on the planet.

We predicted that one day computers would be as fast as they are now, and that they would be able to hold as much info as now. But, we predicted that the computers would be as tall as entire buildings.

AI definitely has a long way to go, but I believe that one day it will definitely surpass us. I also have reason to think that we will not have to fear these intelligent things. We made them, therefore they will respect us. Is it not the duty of a mother to see that her kids grow up to be better than she was? And is it not the duty of the kids to take care of their mother?

 

[CFAlpha]

The ALPHA program at the school I'm attending has a class that attempts to define intelligence, and what it means to be intelligent, gifted, and talented. Invariably, each class ends up with different definitions each year, and many times our definitions change throughout the years. For instance, when faced with the idea that someone may not be very school-smart, but is very street-smart, can't we still say they're intelligent, just in a different way? the same holds true if they're very talented at music or art, yet do not fare well in classes. Maybe they are just intelligent in a different way.
So, measuring intelligence, no matter how you do it, will always raise some sort of controversy. I prefer to think along the lines of "Is it adaptable?" If given an unrecognized command, can it think and find out a way to interpret it, and store how it interpreted it for later? I think once we can create an AI that can learn, then we can worry about measuring it's intelligence.
Also, on the though of multi-tasking: I was taught that no, we cannot multi-task, we're like a single processor, we don't multi-task as much as switch our focus between different tasks very quickly, so it seems as though we're multitasking. I haven't met someone who can think two different things at once. Though, that isn't to say my teacher was correct. That is something I will look into.

 

[The_Magistrate]

No computer can be 'intelligent' in the same ways that humans are intelligent. The brain has a mental capacity that is far beyond what a computer can do. The difference is that computers are very methodical and precise, while humans are prone to errors and tire easily.

There are many ways to make computers intelligent. A couple examples are: genetic algorithms, neural-nets, expert systems... etc.

Universally accepted as the base for testing a computers intelligence is the Turing Test, created by Alan Turing in 1950. It explains that a machine is only truely intelligent when it cannot by distinguished as a computer from another human being. A user (Person A) is sitting in front of a terminal which is connected to another human (Person B) at another terminal and a computer. Person A asks questions of Person B and the computer. If they cannot reasonably determine which user is the computer and which is the human, the computer is deemed artificially intelligent.

 

[oblivionguardiansaints]

Anyone remember Eliza?

I'd assume not.


I've programmed several AI... Yet I've also had to terminate some of them, because it wouldn't follow its programming correctly. (Stupid way to put it, long story.)

I love to see AI working, especially when it does when you want it to do.

But, they'll never be near as smart as humans, even if they can process tons of information within seconds. They firstly won't have emotion, whatsoever. Secondly, logical reasoning would not be very well maintained.
So yeah..

 

[intheclutch]

I think by its very nature, it's impossible to make something smarter than us. It might be able to perform better at some things like calculations, but it won't be able to be smarter than humans because it was made by humans.

 

[taekwondokid42]

I still disagree. I know that it is a very short time until computers have more power than humans, and I believe that one day we will achieve a program that will enable the computer to solve problems and come up with theries and what-ifs based on what it already knows.

But that could be several decades from now.

 

[intheclutch]

I still disagree. I know that it is a very short time until computers have more power than humans, and I believe that one day we will achieve a program that will enable the computer to solve problems and come up with theries and what-ifs based on what it already knows.

But that could be several decades from now.

They have those, computer weather models, as an example. It's a what-if based on what it "knows" (data that it is given). They are not perfect by any stretch of the imagination, but they do equations faster than any human, so that is a plus, but humans still need to figure out of what the computer is saying is valid or if it's junk. That's what I believe we have, and will always have...more efficiency with computers compared to humans, but still not "smarter" than humans.

 

[faolan]

I would also disagree that we would never be able to make something as intelligent as us. At this time, the technology levels and knowledge levels that we have make a *true* AI a near impossibility.

That said, with the way technology is increasing, as well as our increasing knowledge of biological thought patterns/neural mapping, it is only a matter of time before we can translate that directly to electronic data storage and processing algorithms. Indeed, we may wind up having to combine the current data systems and some form of 'wetware' system... in effect creating a biological brain out of tissue for the computer to have.

As regards emotion vs. logic and soul vs. souless... well, there we're getting into philosophy and the whole 'Penfield Vat' vs. 'Berkley World' argument. What is the soul? Is it our thoughts, or what? Of course, we don't even want to start *that* argument on this board.

As for measuring the intelligence of a computer.... Well, The_Magistrate was absolutely right about the Turing test. I will state, however, that if we cannot truly and accurately judge the intelligence of another human across social and cultural boundries (all IQ tests are skewed), then how would we be able to do the same for a new form of intelligence?

 

[lpeters]

I believe that no computers aren't intelligent; I don't believe that the Computer will ever reach the same level as the Human Brain.

This is simply because although Computer Technology is advancing rapidly; it will only become as good as a fraction of the Human Brain. We create these ChatBots through past experiences and how we believe things to be. Different people's brains are developed at different levels and therefore no matter how good we make this technology it will always be held back by our brain.

So I suppose the real answer to the question "is there such thing as Artificial Intelligence" is no purely because the computer is no better than the Human Brain and never will be. I personally sometimes use a ChatBot for when I am away from the computer and when I read the log of what has been said I can see many times when the ChatBot could not answer a question asked. A ChatBot is only as good as it's creator. Therefore it cannot pass our level of intelligence and all it really does is uses a series of if and else to see if it finds anything in a sentance from a corresponding person and if it finds nothing it will just grab a random line and send it. All this really is is Human Intelligence stored in a script.