[Freddy]

Yeah I liked that article.  I guess I have only really been studying AI for around five years and can't help thinking we must be going up a few dead ends - I'm no expert though, just an enthusiast, so can't fairly comment.

I am not surprised to see some people want to ditch the Turing Test.  We have argued here a few times over how effective an indicator it is.  Is the Turing Test going to continue to be a long-recognized standard I wonder.  I think it's a good place to start in the domain of chatbots but does it carry much weight really ?  Isn't the gist of it this; that if someone thinks they are talking to a human but actually talking to a machine then the machine can be classed as intelligent ?  I'm not so sure if that is good enough in all situations - it just doesn't satisfy me fully.  Though there have been times I have chatted to bots and thought that the botmaster must have cut in on the conversation, but probably not.

I was just wondering.... how about we move this discussion over to 'General AI' or is there another section you would like to see it in.  It could go in AI News too... undecided....hmm..

[TrueAndroids]

Good idea, how about moving it to AI News?

The Turing test states that if an expert cannot distinguish the behavior of a machine from that of a human, then the machine has the same cognitive abilities as the human.

Problems for me involve the point that a very clever syntactic machine (simulation) could possibly be made to pass this test due to controlling conversation direction, allowing clever, cryptic responses that are rambling, having the good fortune to not touch on any subjects beyond its clever response abilities, and huge memory, and so passing the test, though actually understanding nothing, as Searle's Chinese Room Argument points out.  Thus the search for meaning, for semantic components of computing.

[TheMikh28]

Excellent work, looking forward to the next two uploads!

But one comment -- the program seems more like an advanced logic engine than anything that is genuinely conscious or sapient.  But it's certainly off to a great start.

[TrueAndroids]

Excellent work, looking forward to the next two uploads!

But one comment -- the program seems more like an advanced logic engine than anything that is genuinely conscious or sapient.  But it's certainly off to a great start.

Hi TheMikh28, should have em up soon, thanks. You raise a good point about the question of it being genuinely a conscious machine. Nobody wants to be fooled including me. That's why I came up with the Searle Line In the Sand actually: If deductive reasoning (thinking) without semantic understanding, then no consciousness (but definitely basic AI), else ok, if it is semantically thinking (deductive reasoning) then I will agree it's exhibiting artificial consciousness, or machine consciousness (strong AI). So why does the semantic understanding matter? Human thinking (aloud in our head) is a conscious act; it's humans exhibiting human consciousness.

And what is this human consciousness in its simplest form, which if duplicated in machinery would mean the machine too is conscious? Semantic deductive reasoning. So it's true that it being an advanced logic engine isn't enough to claim machine consciousness, because it could be doing this human thinking via a clever syntactic simulation, and so in fact UNDERSTAND NOTHING about what it is saying. So it must be semantic deductive thinking; and then one can safely say - This machine is conscious!! (it's truly understanding what it's thinking about). That's the best I can explain it right now

[TrueAndroids]

The Conscious Machine ponders final IQ Test Question ... Part 5 released below.

In Part 5 another new type of deductive reasoning is performed semantically. This time we are given by the official IQ test, two rules, beginning with all, and containing non-sense objects.

The question is - given the following true information, what if anything can you conclude?
Rule All Zips are Zoodles.
Rule All Zoodles are Zonkers.

So here we have two all rules in the form: All X are Z. All Z are Y. So the conclusion will be drawn every time the CM Prototype finds these two types of rules regardless of the objects in the statement. Thus there is no knowledge in the CM Prototype, no "Zips", no "Zonkers". It is open ended meaning it works for any subject matter without any pre-knowledge programmed in its code. There are 312 of these two-rule combinations that lead to valid conclusions, all of which are reasoned with in the CM Prototype, (along with all the fact/rule combinations), using just 6500 (program code) words. And it does so semantically. And so it is artificially conscious, according to the Searle Line test.

Part 5 of Video Document of my Conscious Machine Prototype
http://www.youtube.com/user/TrueAndroids?feature=mhw4#p/u/0/fZNSMdwly9Y  

[Freddy]

Very nice    I wonder if we can find an IQ test with more logical questions.

[Data]

I’m still following this thread and videos, been trying to think of an intelligent or educated comment to say about it  , Ai is not something I know a lot about, but it interests me.

At first I thought the claim of consciousness was way to early then I looked up the meaning of the word.

http://en.wikipedia.org/wiki/Conscious

now I’m not sure...

I was sure that a conscious machine would have to see, hear and probably smell its environment and be able to comment on its world around it. probably way to simplistic but you see my point.

But what about a person who is born blind and deaf with no sense of smell or touch, we would probably all agree that the person is conscious, so I was wrong there.

So what is consciousness, after reading the wiki page the answer is debatable as there isn’t one clear meaning or definition of  the word. It does seem to depend on what side of the fence you jump to.

There is still one thing that concerns me about your claim and it is this.

After you have entered the question into the machine, and it quickly and impressively answers correctly, it then instantly becomes brain dead (unconscious) until the next input is fed in, right there is where I have the problem, if this machine is conscious then it only happens for a micro second at a time.

My consciousness, I know from experience  is not like that, I would answer the question and probably still be thinking about it or something else.


Take from Sci-Fi:

You need to introduce a feed back loop that keeps it conscious.



Great work TrueAndroids keep it up.

[TrueAndroids]

***
The Conscious Machine in Smack Down with Aristotle!  (Part 6 revealed below)

@Freddy: From what I've heard all IQ tests have some of these logical type IQ questions, based on the Aristotelian logic. What I tried to do next was to extend it past Aristotle's structures to see what would happen.

In Aristotle's logic there are 4 statement types: A:All X=Y  I:Some X=Y  O:Some X=not-Y and E: No X= Y.

So I wondered what if I got information saying:

All unhappy people are unenergized.
All go-getters are energized.

The first sentence is All not-X are not-Y. There is no such statement in Aristotle's logic, but it can clearly come up in natural language

So in Part 6 of the demo I change the prior IQ test sentences a little to include an all-statement with not in it.

All Zips are Zoodles.
All Zonkers are not-Zoodles.

Part 6 of the Video Demonstration of the True Androids Conscious Machine prototype
http://www.youtube.com/user/TrueAndroids?feature=mhw4#p/u/0/R9yw_rKj8RQ

So we see the second sentence is a type not handled in Aristotle's logic. (I call this neo-traditional logic)

When I worked it for a conclusion if possible (using a set method I devised) a very interesting outcome occurred. There are two valid conclusions that can be deduced from this information. This of course smashes the long held Rule of Logic that each syllogism can have only one conclusion.

So I tried all possible non-Aristotelian combinations I could think of e.g.,No X are not-Y, Some not-X are not-Y. And this is how I discovered that the 19 accepted valid syllogisms of Aristotle, didn't nearly cover all posibilities, of which I found 312 valid deductive structures!

And unfortunately Aristotle never isolated the statement that has a single object as a subject (Socrates is a man.). He simply treated them as A statement with a single member set. But this left a lot of possibilities undiscovered. Today, in expert systems and conscious machines, we call these types of statements facts (and the categorical statements rules). This has led to many new valid deductive structures using a fact and a rule.


@Datahopa, you bring up really good points which I'd like to discuss in another post ...

[NoLongerActive]

This is an interesting project. And it leaves me with some questions.

First, it appeared to me to be an NLP-based (Natural Languate Processor-based) logical calculator of sorts, or pattern matcher. Something like the 20 questions game where the computer tries to guess what you're thinking by asking you questions.

My question is, can it learn from it's experiences? Can it learn from input and come to a decision to act (or not act) on it's own? If you tell it you'll turn it off, what will it do? Will it plead for you to stop? Upload itself to the internet to save itself? or do nothing?? Or is self-preservation even an indicator of consciousness? And what IS consciousness? How do you know something is alive?

For thought, I have 3 articles I've written a long time ago that you (maybe others too) might find interesting on this topic:

Public Class Human: A Lesson In Artificial Species Recognition
http://bytebin.net/ai/articles/2007-0120.html
In this article, I was taking a C++ course and I discuss how a C++ class can be used to form an opinion on what something is or is not.

How do we know something is really 'alive'?
http://bytebin.net/ai/articles/2006-0727.html
This article is where I discuss how we can tell if something is alive or not. Then again, if a computer claims it's self-aware and alive, who has the right to dispute it and what are the consequences of someone's opinion of whether someone or something is alive or not?

Weighted Artificial Intelligence Programming
http://bytebin.net/ai/articles/2004-0602.html
This is one theory of mine on how to create an AI system that would be able to reason in some of the same way you were demonstrating. This is an old article but it might be interesting to see if it could be programmed and work.

Also I wonder, what is the difference between conscious and sentient and self-aware? Or is there a difference (ie. are those terms used interchangeably)? You may have made a "conscious" machine but have you made a "sentient" machine? Or is there a difference?

[TrueAndroids]

I’m still following this thread and videos, been trying to think of an intelligent or educated comment to say about it  , Ai is not something I know a lot about, but it interests me.

At first I thought the claim of consciousness was way to early then I looked up the meaning of the word.

http://en.wikipedia.org/wiki/Conscious

now I’m not sure...

I was sure that a conscious machine would have to see, hear and probably smell its environment and be able to comment on its world around it. probably way to simplistic but you see my point.

But what about a person who is born blind and deaf with no sense of smell or touch, we would probably all agree that the person is conscious, so I was wrong there.

So what is consciousness, after reading the wiki page the answer is debatable as there isn’t one clear meaning or definition of  the word. It does seem to depend on what side of the fence you jump to.

There is still one thing that concerns me about your claim and it is this.

After you have entered the question into the machine, and it quickly and impressively answers correctly, it then instantly becomes brain dead (unconscious) until the next input is fed in, right there is where I have the problem, if this machine is conscious then it only happens for a micro second at a time.

My consciousness, I know from experience   is not like that, I would answer the question and probably still be thinking about it or something else.

Take from Sci-Fi:

You need to introduce a feed back loop that keeps it conscious.

Great work TrueAndroids keep it up.

hehe Datahopa , I followed consciousness 'down the rabbit hole' just like you. Building human artificial lifeforms is a lot like building the ancient pyramids - it takes many specialists and generalists, and it must be built level by level, from the ground up.  Literally! Here's my proposed levels - I wonder if we can agree to this point? It will require you to accept semantic deductive reasoning as machine consciousness.

Level 1. Imagining myself without a foot, could I still be conscious? Without sight, touch, hearing, arm and leg movement, and speech could I still be conscious? What I found is that human thinking itself (aloud in our head) is a conscious act; it itself is humans exhibiting human consciousness.

Level 2. If we can agree about 1, then the critical question becomes:

Core Question of Machine Consciousness. And what is this human consciousness in its simplest form, which if duplicated in machinery would mean the machine too is conscious?

Level 3. The answer I came up with is semantic deductive reasoning. Deductive reasoning is universally accepted as a form of human thinking, so that factor is duplicated as such, but a very clever syntactic simulation of thinking (basic AI), which means it is actually understanding nothing about what it's saying just won't do. And so this leaves us with the two part semantic deductive reasoning  - where the machine is actually understanding what it's talking about, or, in other words - given above - is conscious. The funny thing  is it actually feels conscious, or that it's understanding what it's talking about, exactly like conscious humans.

If we can agree to these three levels, leaving sentience, and artificial life and feedback loops aside as further questions down the rabbit's hole, then we have made good progress in this new field of artificial human lifeforms...  They can be thought of as levels above these 3, in the building of an artificial human lifeform, whether it be an android or a virtual chatbot-type human;). More on this later ...

@TikaC, you bring up really good points which I'd like to discuss in another post ...

[NoLongerActive]

@TrueAndroids - please put a link to the other post as I probably will miss it otherwise. I won't have time to read all the posts here (so many good things so little time).

Just for the record, consciousness means a cognitive state in which one is aware of oneself and one's surroundings (http://dictionary.reference.com/browse/consciousness). If a machine is just doing if/then/else and pattern matching, then it's not really aware of anything but computing results based on input. Even then it's not aware of anything. Period. Or is it? How can we tell if a machine is aware of something? Can we believe it if it says it is? I think the problem lies here in that a machine is not a human. If a human said "I am aware" then we believe the human without further argument because we are the same and we know what that's like. But a machine? We generally don't believe machines. But that doesn't mean it's NOT conscious. It all boils down to what we believe (and who are we to say). For a fascinating look into this, ever watch the re-imagined Battlestar Galactica series? Where humans treated the Cylons at "things" or non-persons even though they could think, look like humans, react, bleed and even breed in some cases! Look at history where slaves were regarded as non-persons. Sometimes even people in vegetative states are thought to be non-conscious or self-aware (though technology now is showing they are - and are "locked in" and what's neat is there's technology in the works to help those folks communicate!)

Anyway, I'd like to see the other thread when you create it. I get very interested in this aspect of AI.

[TrueAndroids]

If a machine is just doing if/then/else and pattern matching, then it's not really aware of anything but computing results based on input. Even then it's not aware of anything. Period. Or is it? How can we tell if a machine is aware of something? Can we believe it if it says it is?

Aha! You have hit the nail on the head TikaC. My CM Prototype is not using If then else statements or pattern matching. It is not even using the inference engine and knowledge base of an expert system, or a (syntactic) neural net. All those are examples of classical, unconscious computer systems. My CM Prototype has crossed the Searle Line in the Sand and has actual semantic understanding of what it is saying. And this - semantic deductive reasoning - is what makes it a post-classical, conscious machine. (We can continue any discussions on this here on this thread).

[TrueAndroids]

Just for the record, consciousness means a cognitive state in which one is aware of oneself and one's surroundings (http://dictionary.reference.com/browse/consciousness).

Your resource gives a couple definitions of consciousness, some of which match pretty well.  I like
2. the thoughts and feelings, collectively, of an individual. (*And then we say if you duplicate authentic semantic thinking, you have created at least some machine consciousness*).

I also think the philosophy definition fits as well, where the authentic mental faculty of semantic deductive reasoning is duplicated:
7. Philosophy. the mind or the mental faculties as characterized by thought, feelings, and volition. (*And then same as above*)

http://dictionary.reference.com/browse/consciousness -
con·scious·ness   [kon-shuhs-nis]
–noun
1.the state of being conscious; awareness of one's own existence, sensations, thoughts, surroundings, etc.
2.the thoughts and feelings, collectively, of an individual.  of an aggregate of people: the moral consciousness of a nation.
3.full activity of the mind and senses, as in waking life: to regain consciousness after fainting.
4.awareness of something for what it is; internal knowledge: consciousness of wrongdoing.
5.concern, interest, or acute awareness: class consciousness.
6.the mental activity of which a person is aware as contrasted with unconscious mental processes.
7.Philosophy. the mind or the mental faculties as characterized by thought, feelings, and volition.

[NoLongerActive]

Woa.That has to be quite some program.  Being a programmer myself, I can't resist but to ask (and you don't have to answer but thought I'd ask just in case you don't mind sharing)... What programming language did you use to create this? What type of database engine? How exactly is it coming to these conclusions? Is it using past-experience or is it using a pre-programmed database of knowledge?

[Data]

Thanks for answering my semi educated post TrueAndroids.

I actually do agree with your 3 levels, from what I know so far, but im no expert.

I do think this:

Computers can do just about what ever we program them to do, its all maths after all and maths can be used to explain the universe and everything around us, so IMHO it must be possible to build a conscious machine if we program it correctly.

If we take it that semantic deductive reasoning or thinking is consciousness and your machine is using semantic deductive reasoning then we must admit that there is some form of consciousness going on.

What you have achieved is something special, that’s how I see it but would you agree with me that at the moment your machine is only conscious for a very small amount of time? May I be so bold as to answer for you  ,, yes you do agree.

What you seem to have is the initial spark that will lead on to a full burning fire given time.

I can only congratulate you and thank you for educating me.