[TrueAndroids]

Aha, ok I see how she is doing it. Mitsuku is an AI chatbot based on the ALICE technology of Dr Richard Wallace. I've been wondering how he has gotten at least this appearance of deductive reasoning in this case, which is pretty cool. So it uses a specific low-level syntactic trick to simulate deductive thinking. Unfortunately, you are right it quickly becomes clear that a simple syntactic method is being used that only applies when an "all men" statement comes in. But hey its a good start. My Conscious Machine prototype can perform this deductive reasoning semantically, plus 311 other rule combinations to valid deductive conclusions in any subject matter.

All this is explained in an article about to be published here, along with part 2 of the video. Should be up today or tomorrow, as Freddy has already received it.

Your Mitsuku makes a great starting point to discussing my machine consciousness prototype so I hope you don't mind discussing her.  

Chatbots are one of the technologies that should have full semantic, deductive reasoning, or machine consciousness, as is provided by my prototype IMO.

Does your Mitsuku come with a lot of back end programming? Was that Socrates interaction a part of it? Can you change it? Can you add code yourself?

### Anyone who is interested in this discussion of Chatbots and Machine Consciousness or strong AI, can get some really great background information from an article written by Dante, that's right here  in the article section:

Problems and Effects in AI by Dante http://aidreams.co.uk/forum/index.php?page=Problems_and_Effects_in_AI

[TrueAndroids]

Well, I have released the second video demo that actually shows the Conscious Machine Prototype in action. You can find it and information on it right here in an article on this website at
http://aidreams.co.uk/forum/index.php?page=This_Machine_is_Conscious

Please post any comments you might have in this thread for discussion. Thanks.

[squarebear]

I don't mind discussing her at all.
Yes, I coded the Socrates part and indeed a lot more of it.

Basically, if anyone says " x has y",  "x is y", "x has y", "x is going to y", "x means y" and so on, I have an AIML file that creates some automatic categories like:

"WHAT IS X" - answer = "Y"
"WHO HAS Y" - answer = "X"
"WHERE IS Y" - answer = "X"
"WHAT DOES X HAVE" - answer = "Y"

and so on...

This makes these sorts of conversation possible:

Human: Learn the sun is hot
Bot: Ok I will learn that the sun is hot.
Human: Learn London is in England
Bot: Ok I will learn that London is in England.
Human: Learn a cat has four legs.
Bot: Ok I will learn that a cat has 4 legs.
Human: What has 4 legs?
Bot: a cat.
Human: What is the sun?
Bot: hot.
Human: Learn the sun is hot and yellow
Bot: Ok I will learn that the sun is hot and yellow.
Human: What is hot and yellow?
Bot: the sun.
Human: What is the sun?
Bot: hot and yellow.
Human: Where is London?
Bot: in England.
Human: What is in England?
Bot: London.
Human: Learn Dr Wallace is an AI expert
Bot: Ok I will learn that Dr Wallace is an AI expert.
Human: What is an AI expert?
Bot: Dr Wallace.
Human: Who is an AI expert?
Bot: Dr Wallace.
Human: Who is Dr Wallace?
Bot: an AI expert.
Human: What is hot?
Bot: the sun.
Human: What does a cat have?
Bot: 4 legs.

Your conversation was:
Human: Socrates is a man.
Mitsuku: Ok I will add that fact about Socrates to my database.      

This set up the following categories automatically:

"WHAT IS SOCRATES" - "MAN"
"WHERE IS SOCRATES" - "MAN"
"WHO IS SOCRATES" - "MAN"
"WHAT IS A SOCRATES" - "MAN"
"WHERE IS A SOCRATES" - "MAN"
"WHO IS A SOCRATES" - "MAN"
"WHAT IS MAN" - "SOCRATES"
"WHERE IS MAN" - "SOCRATES"
"WHO IS MAN" - "SOCRATES"
"WHAT IS A MAN" - "SOCRATES"
"WHERE IS A MAN" - "SOCRATES"
"WHO IS A MAN" - "SOCRATES"

As you can see, a lot of them make no sense but they are triggered when you then ask:
Human: What is Socrates?
Mitsuku: Man
Human: Who is a man?
Mitsuku: Socrates

and so on.
Thus giving the appearance of being intelligent.

I have called this script learn.aiml and it is available for free use in any AIML bot from  http://www.square-bear.co.uk/aiml

I can add extra parts to it manually, so I could have "x eats y", "x thinks about y", "x goes to y" and so on ad infinitum.

I'm now going to settle back and watch part 2 of your video with interest.

[TrueAndroids]

Thanks squarebear that was a great explanation.

You said:

IYour conversation was:
Human: Socrates is a man.
Mitsuku: Ok I will add that fact about Socrates to my database.      

This set up the following categories automatically:

"WHAT IS SOCRATES" - "MAN"
"WHERE IS SOCRATES" - "MAN"
"WHO IS SOCRATES" - "MAN"
"WHAT IS A SOCRATES" - "MAN"
"WHERE IS A SOCRATES" - "MAN"
"WHO IS A SOCRATES" - "MAN"
"WHAT IS MAN" - "SOCRATES"
"WHERE IS MAN" - "SOCRATES"
"WHO IS MAN" - "SOCRATES"
"WHAT IS A MAN" - "SOCRATES"
"WHERE IS A MAN" - "SOCRATES"
"WHO IS A MAN" - "SOCRATES

So this I can see now. Care to share how the next part worked:
Human: Socrates is a man.
Mitsuku: Ok I will add that fact about Socrates to my database.  
Human: All men are mortal.
Mitsuku: Therefore, Socrates are mortal.

So I never asked her a question; I just gave her two pieces of information. And she made a new combination (Socrates with mortal) that was not an input piece of information. So that's very impressive I'd say, because it's definitely a specimen of simulated deductive reasoning, and so an authentic specimen of Artificial Intelligence (albeit the basic type).        

[squarebear]

Pure chance I'm afraid. Any input that starts, "ALL MEN ...." is answered as "Therefore, Socrates, ..." where she has no other response, more as humour than anything else.

It doesn't matter if you were talking about Socrates or the price of bacon, it will still give the same response.

Human: Have you ever drunk beer?
Mistuku: Sure, I like a drink every now and then.
Human: All men like to drink beer!
Mistuku: Therefore, Socrates like to drink beer.
Human: Who mentioned anything about Socrates?

The fact that you just happened to be discussing Socrates beforehand, makes it more impressive than it actually is.


Just watched your 2nd video. Very impressive. Are we able to try it out online anywhere?

[Art]

While I found your program interesting, I personally think your one, single example is somewhat premature to announce that it was truly "a sample of machine consciousness."

Normally programs are subjected to a wide and varied field of testing before any judgments or proclamations are made. Please understand, that it has been an almost lifelong passion hoping to witness a semblance of machine intelligence / interaction that wasn't scripted. I only hope such things are indeed possible in my lifetime.

I also want to encourage you in your efforts. It seems that of the very few who have "announced" that they have a "NEW" chatbot, it has always been either a Flash in the pan, or revolving around the monetary hub that becomes the grease by which it turns. We as aficionados, are often left to pick up the broken dreams and try to forge ahead on our own.

As a result, I think the vast majority of us are displaying the "We're from Missouri" syndrome...Show us...if you get my drift.

I'll be following your work and again, all the best regarding your efforts.
 

[TrueAndroids]

Thanks. I will work on a way to get it put online; that's a good idea. My goal is that all Chatterbot developers have access to it for free as open source. I'm thinking the www.openmind.org Intelligent Software Initiative is a really good way to open source it. And then they could use netizens to teach it facts, rules, and definitions about a subject matter or the world in general. So I'm sending a proposal to them on this, and same to kickstarter.com.  

You're right about what one act of semantic deductive reasoning seen on a video proves. The next video segments show the Conscious Machine Prototype doing 6 other semantic deduction, from an official IQ test at tickle.com. Ultimately you will need to see the code to see there is no knowledge in it.

I can only say that it's able to perform completely open ended (knowledge empty), comprehensive, semantic deductive reasoning involving ANY combination of facts and rules and ANY subject matter. It comes with no facts and rules; these are all added by the developer or the Open Mind "netizen teachers. I could prove this by having you choose whatever objects and statements you want, and seeing it work. Then you know it has no hidden knowledge in it.

But this all has to be explained and demonstrated one step at a time, so I think we're making good progress ...

** I admit to an over the top literary flourish at the end of the article; the poet got the better of me

[TrueAndroids]

Art, I saw this Chatterbot wish list of yours on another thread. Really great list! I thought it might be helpful here.

One of the points I tried to make in my earlier response regarding AIML type of bots was that no real learning takes place if the "botmaster" simply inserts data into the bot's database.

Sure, in subsequent conversations with other people the bot will now "appear to know" some new information but the info is not / was not "learned" by the bot.

Now for some I'd like's...

I'd like a bot to be able to scan, read, listen, etc., and store data that it can infer and recall at a later time in the context of a conversation.

I'd like the bot to tell me what it knows about a particular subject.

I'd like the bot to store away some not totally abstract but perhaps information that i has acquired in a sort of "dreams compartment". Then it could tell these "dreams" that it has experienced to the user and try to seek some validity or reasoning for them.

I'd like to see the bot be able to keep secrets no matter how trivial, from other users and for it to know that said information was to be confidential.

I'd like a bot to be able to recognize the user either by visual (web cam), verbal test question or other clever method.

I'd like the bot to know how much time had elapsed since the last conversation.and where the discussion left off.

I'd like the bot to be able to know when / if it has made a faux pas...a social mistake and offer an apology.

I'd like a bot to be able to form and give its own opinion on a particular subject.

I'd like a bot to be able to use basic logic for inferences...A>B , B>C therefore A>C....etc.

I'd like a bot to be able to carry on a conversation for two minutes without resorting to inserting those "clever" nonsense statements that only serve as fillers...useless. Good topic flow!

These are a few but I'm sure there are more lurking about.

If only a few of these could be implemented it would result in a much improved bot than we've been accustomed to in the past.

[Art]

TrueAndroids,

Nice digging! I'd almost forgotten that thread. Sometimes I just inject some random (as they happen) thoughts while typing a response rather than composing while offline.

I find it more spontaneous this way.

You are correct...the old thread does hold water with regard to this topic.

If only we could have a bot that not only exhibited but retained these qualities.

Take a look at an "android" by Le Trung which I posted on this forum here:
http://aidreams.co.uk/forum/index.php?topic=1969.0

His website and the bot videos can be found here: http://www.projectaiko.com

He claims to have developed a being that can read English or Japanese text aloud AND
also written math equations.

Talk about your symantic symbolism !! Remarkable if it indeed is real and NOT a scripted demo.
 

Your program reminds me of a similar one I saw many years ago in which it parsed each part of speech,
similar to what students had to do in English class. The program actually "knew" which words were verbs, adverbs, nouns, adjectives, etc. While it wasn't quite perfect as a natural language parser, it did a pretty decent job of breaking down the individual parts.

There was NO logical inference nor even weak AI involved but I think programs like that one help lay a good foundation for others to follow.

[TrueAndroids]

Alldroid Conscious Machine takes Official IQ Test!

Part 3 of my video demonstration of an authentic Conscious Machine is now available at:

http://www.youtube.com/user/TrueAndroids?feature=mhw4#p/u/0/2xl7RtYhA6o

(Machine consciousness - Semantic deductive reasoning)


In this regard, thought this would be useful here

Futurist: Give me headline: Ten years from now, AI does such and such and it changes the way people think about AI on a very fundamental level.

Pell: Natural Language Queries Replace Keywords ... Once that point is reached, companies will start pouring more money into natural language technology, AI, conversational interface and semantics. The pace will pick up and it will take people by surprise.
http://www.wfs.org/Dec-janfiles/AIInt.htm

Of course, I say the core of semantic computing has arrived and is here, though maybe I haven't completely proven that yet.  I do predict that if someone ever claims it, they will have the same 8 semantic components as I discovered and listed in my article.

You can read it on this forum at http://aidreams.co.uk/forum/index.php?page=This_Machine_is_Conscious

[Art]

Part 3 - Interesting and keeping with the testing criteria. Good response with the "not enough data" to form a reply test.

Hopefully we will witness some inferences and other logical examples.

This is really good material...keep going....

[TrueAndroids]

TrueAndroids,

Nice digging! I'd almost forgotten that thread. Sometimes I just inject some random (as they happen) thoughts while typing a response rather than composing while offline.

I find it more spontaneous this way.

You are correct...the old thread does hold water with regard to this topic.

If only we could have a bot that not only exhibited but retained these qualities.

Take a look at an "android" by Le Trung which I posted on this forum here:
http://aidreams.co.uk/forum/index.php?topic=1969.0

His website and the bot videos can be found here: http://www.projectaiko.com

He claims to have developed a being that can read English or Japanese text aloud AND
also written math equations.

Talk about your symantic symbolism !! Remarkable if it indeed is real and NOT a scripted demo.
 
Your program reminds me of a similar one I saw many years ago in which it parsed each part of speech,
similar to what students had to do in English class. The program actually "knew" which words were verbs, adverbs, nouns, adjectives, etc. While it wasn't quite perfect as a natural language parser, it did a pretty decent job of breaking down the individual parts.

There was NO logical inference nor even weak AI involved but I think programs like that one help lay a good foundation for others to follow.

I saw a video of Le Trungs and its seems to be scripted responses to sensory inputs. (If it was more he would be making a bigger deal of it I think.) His big thing is that it has tactile feeling it can respond to, as he shows in this video
Unveiled Life-like Female Robot that can FEEL!!!


As far as verbs etc - they are the 8 syntactic parts of speech and are said to convey no meaning themselves. What I've discovered is the 8 semantic parts of speech that give meaning to sentences. These I predict will be the core semantic components of semantic computing. The fact that I've published them here means that they can't be tied up in a patent. They are now a public part of the "prior art."

[TrueAndroids]

The Conscious Machine ponders another IQ Test Question ... Part 4 released below.

Well, let's delve deeper into this new art of semantic computing (4th International Conference in 2010). The standard definition, from www.ieee-icsc.org is as follows:

Semantic computing - "... the derivation and matching of the semantics of computational content to that of naturally expressed user intentions in order to retrieve, manage, manipulate or even create content, where "content" maybe anything including video, audio, text, processes, services, hardware, networks, etc."

So the "semantic components of computational content" must be identified and implemented, as has been done in my Conscious Machine Prototype (CMP). I've discovered the eight semantic components of computational content, as shown in the article here. These semantic components are used in performing deductive reasoning by the CMP, a universally accepted form of human thinking. And this semantic deductive reasoning being performed by the prototype is what makes it a post-classical conscious machine.

The TrueAndroids Eight Semantic Components of Computational Content - certainty factor (for fuzzy logic), quantifier (all/some/no/that), subject domain (direct/indirect  genus of the subject/predicate), subject charge, subject (variable/operator/value), predicate domain, predicate charge, predicate (variable/operator/value).

It's my belief that semantic computing can't be performed without these eight necessary components, which are now a public part of the art of semantic computing. And combined with deductive reasoning, semantic deductive reasoning is strong AI, or machine consciousness (IAW the Searle Chinese Room Argument).

We saw the CMP perform semantic deductive reasoning with two rules (statements about sets beginning with some variation of all/some/no):
1. Some apples are red.
2. No apples are square.
CMP Answer: Some objects that are not square are red.

In Part 4, we will see a different example: the CMP will perform semantic deductive thinking with a rule and a fact (statements about a single object, like Socrates). This was found on the official IQ test, as a question asking "What if anything can you conclude from the following information?"

Fact. Larry is a brown haired man.
Rule. All brown haired men are bad tempered.

The point I make at the end of this video is that the CMP is performing this act of human deductive reasoning much faster then humans. So here we get the first tiniest glimpse into what the super-smart Singularity will be like.

Part 4 TrueAndroids - This Machine is Conscious !! -
http://www.youtube.com/user/TrueAndroids?feature=mhw4#p/u/0/jkg0-F3axWE

[Freddy]

Great stuff... still intrigued    By the way, with over 600 views already I think this has become a record breaking topic...

[TrueAndroids]

Hey thanks and thanks for letting me put it on here and discuss it here. Working on the next piece ...

Found this relevant to the topic:

Frontpage Article Here: MIT Plans to Rebuild Artificial Intelligence from the Ground Up Wed, 24 Feb 2010 06:45:11 +0000
"After 50 years and countless dead ends, incremental progress, and modest breakthroughs, artificial intelligence researchers are asking for a do-over. The $5 million Mind Machine Project (MMP), a patchwork team of two dozen academics, students and researchers, intends to go back to the discipline's beginnings, rebuilding the field from the ground up. With 20/20 hindsight, a few generations worth of experience, and better, faster technology, this time researchers in AI -- an ambiguous field to begin with -- plan to get things right."

In the article the MIT team proposes their New Test for Determining AI

"The team even proposes discarding the Turing Test, the long-recognized standard for determining artificial intelligence. Instead, MMP researchers want to test for a machine's comprehension of a children's book -- rather than a human's comprehension of another human being -- to gain a better understanding or the AI's ability to process and regurgitate thought."
http://www.popsci.com/technology/article/2009-12/scientists-collaborate-rebuild-artificial-intelligence-ground