We get clowns passing through here on an almost weekly basis.

True, but that’s what makes it fun

haha y u so mad cause someone else is striving for greatness and you are not?

Do you know what Steve has done already in the area of chatbot development besides Mitsuku and the chatbot battles?

Genesis,

It’ll run on my version number of Windows
which is ninety version numbers higher than 8,
right?

Do I need to downgrade from Windows 98 to Windows 8?

Merlin - Jul 29, 2012:

Machines may always suffer from the “Wizard of Oz” effect (Don’t mind the man behind the curtain). As they accomplish more intelligent looking actions, they will always be thought of as using “clever programming”. Even if they write that program themselves.

The Paradox of Artificial Intelligence.
An article on the difficulty of creating something called AI.

Steve Worswick - Aug 1, 2012:

Genesis - Aug 1, 2012:

haha y u so mad cause someone else is striving for greatness and you are not?

I guess I’m just allergic to bulls**t. We get clowns like yourself passing through here on an almost weekly basis. People who claim to have had major breakthroughs, yet when we ask to see any sort of demo or examples, suddenly they provide none or say that the programs are in foreign languages or need an opertaing system upgrade.

The last guy reckoned he had a program that could pass the Turing test 100% of the time, which I thought was strange as surely such a major breakthrough would have made the news. It was only when it got down to details, we discovered that he had invented his own test where the judge told the computer what to say!

However, if October comes and passes and people the world over are discussing your vaporware, I will apologise.

Here’s the thing. There are hobbist coming here to talk about their hobby. There are also theorists coming here to talk about their theories. the difference from somone who is making a commercial product is that one has to spend an extraordinary amount of money and time and also prove his application in order to sale it. The others do not. But as they said, its just their hobby or its a thoery they have. They don’t have to prove jack.

Its kinda like the people who say they have a new groundbreaking idea/invention and the people actually working on their groundbreaking idea/invention/business. One is talking abuot how his idea is the best thing since sliced bread while spending 0 dollars and time on actually doing it. While the other is spending all their time plus tens and hundreds of thousands on their idea.

Steve Worswick - Aug 1, 2012:

Do you have any examples of dialog with your program or any examples of what it can do that is so much better than anything anybody else is doing?

I don’t give out examples because when I launch and release my vids I want people jaw dropping and saying “UNREAL”, “take my money”. LOL.

But the word “assistant” has been completely tainted by the likes of gimmicks like siri. Who are actually just information retrieval applications like an ordinary google search. There’s no assisting in answering “who is the president of Iraq?”.

But actual assisting is for example: I’m typing up the first stanza of the star spangled banner in Microsoft Word and I had:

O say can you see by the dawn’s early light,
What so proudly we hailed at the twilight’s last gleaming,
Whose broad stripes and bright stars

and asked “what’s next?” or “what’s after bright stars?”.
It would tell me ” through the perilous fight”

If I had:

O say can you see by the dawn’s early light,
What so proudly we hailed at the twilight’s last gleaming,
Whose broad stripes and bright stars through the perilous fight,
And the rockets’ red glare, the bombs bursting in air,
Gave proof through the night that our flag was still there;
O say does that star-spangled banner yet wave,
O’er the land of the free and the home of the brave?

and asked “What am I missing?” it would answer ” You are missing O’er the ramparts we watched, were so gallantly streaming after perilous fight”.

Again this is just one example and yet this example is something we see happen in everyday life.

What we have today are voice searches not assistants.
Its a shame what a few ads and millions of dollars in marketing can do.

What would you call an extraordinary amount of time and money?

Jan Bogaerts - Aug 6, 2012:

What would you call an extraordinary amount of time and money?

It depends on the situation. Most people don’t understand the gravity of an up-taking this is. Its the reason you see lots of theories and very few working model except ones from universities.

Creating a self developing memory system that starts from absolute scratch and evolves is no joking matter when you actually start working on it. Things are guaranteed to go haywire and leave u back on the drawing board.

Rinse and Repeat.

Written designs are always miles ahead of the actually coding process.

An AI starting completely from scratch unlike Skynet, has to gather input and build its memory off of any pattern it sees, and patterns of that pattern, and pattern of it seeing that pattern…etc. That’s extremely difficult to successfully code. Its so easier to do what Merlin does with Skynet.

Windows RTM is already out to the manufacturers and of-course developers get their hands on it August 15. The past couple days I have been tempted to save myself the time and trouble and put the needed concepts into the memory by hand. Since I want a beta by Sept.

This why my view on True AI has drastically changed. Its also ever changing. But Its irrelevant whether or not AI can think like humans or not. Does Skynet think? Personally, No. Does it matters? Not at all.

What matters is whether the results are good and right now chatbots like skynet are lacking behind.

Genesis - Aug 6, 2012:

An AI starting completely from scratch unlike Skynet, has to gather input and build its memory off of any pattern it sees, and patterns of that pattern, and pattern of it seeing that pattern…etc. That’s extremely difficult to successfully code. Its so easier to do what Merlin does with Skynet.

It is actually easier to write the code that collects patterns and patterns of patterns. Bayes rules and Markov Chains can do that. Mark V Shaney demonstrated that in the 80’s.
It is just harder to direct the results to something that seems to be “intelligent” and requires a huge valid input dataset to be usable. Unigram, Bigram, Trigram… etc. modeling also don’t handle uncertainty as well as some other techniques and there are more efficient ways to code and develop than letting the AI to try to “spontaneously combust” into intelligence. If that’s all it took, Google would win, they make heavy use of Bayes rule and have access to more data than anyone.

Genesis - Aug 6, 2012:

What matters is whether the results are good and right now chatbots like skynet are lacking behind.

I think what you meant is if all that matters are the results then “Skynet-AI is ahead”, since it has bested virtually all others and there are very few chatbots at Skynet’s level.

Merlin - Aug 6, 2012:

It is actually easier to write the code that collects patterns and patterns of patterns. Bayes rules and Markov Chains can do that. Mark V Shaney demonstrated that in the 80’s.
It is just harder to direct the results to something that seems to be “intelligent” and requires a huge valid input dataset to be usable. Unigram, Bigram, Trigram… etc. modeling also don’t handle uncertainty as well as some other techniques and there are more efficient ways to code and develop than letting the AI to try to “spontaneously combust” into intelligence. If that’s all it took, Google would win, they make heavy use of Bayes rule and have access to more data than anyone.

I wanted to make a few edits unfortunately chatbots.org was laggy. One was “That’s extremely difficult to successfully code In such a way that gives rise to intelligence.”

So yeah, I’m familiar with Markov and Bayes. I have also seen a few Markov based chatbot and they are a mess. That’s definitely not my direction. If it were i won’t have such a hard time.  Though its a bit similar.

Merlin - Aug 6, 2012:

I think what you meant is if all that matters are the results then “Skynet-AI is ahead”, since it has bested virtually all others and there are very few chatbots at Skynet’s level.

I meant results compared to the goal not its fellow chatbot competition.

Here is Genesis’ posts reduced through a Markov Chain text generator, no canned responses. But is this thinking?

You get my point.
Thanks for asking.
ME: Where is sally?
SKYNET: sally, where is she?
You missed it.
Imagine going to Google Lab and saying “I have a database where you save and use data like name doesn’t mean your bot is canned.
It would be much easier to do what Merlin does with Skynet.
Other than him I’m not aware of anyone who have seen it.
But just like any other chatbot, it’s a chatbot that is a post and push the submit button.
google “Cognitive Architecture”.

Its kinda like the people who say they have a new groundbreaking idea/invention and the people actually working on it.
But Its irrelevant whether or not it can be chaotic.
Its also ever changing.
How mad are you right now I have on file.
While the other is spending all their time plus tens and hundreds of thousands on their idea.
Why spend so much money marketing something when you change the sentence structure from the most obvious.
I thought there was no way in hell a NLP based system could achieve anything in terms of general intelligence.
Only because i forgot to unlist and privatize my video prototype.
Not at all.
Therefore it can’t understand.

Let me ask you this. Does your AI understand or have the capability to understand the concept of Left.

yes

Yes, mine too appears to understand the different meanings of “left”:

Human: A cube is to the left of a pyramid.
Mitsuku: Ok I will add that fact about cube to my database.
Human: Where is the pyramid?
Mitsuku: It is to the right of a cube.

Human: If I have three apples and John takes one away, how many do I have left?
Mitsuku: That would leave you with two.

Steve Worswick - Aug 7, 2012:

Yes, mine too appears to understand the different meanings of “left”:

Human: A cube is to the left of a pyramid.
Mitsuku: Ok I will add that fact about cube to my database.
Human: Where is the pyramid?
Mitsuku: It is to the right of a cube.

Actually that is very good.

But…

Human: A cube is to the left of a pyramid
Mitsuku:  Ok I will add that fact about cube to my database.   
Human: A cone is to the right of a cube
Mitsuku:  Ok I will add that fact about cube to my database.   
Human: what is between a cube and a pyramid
Mitsuku: That which is confined by a cube and a pyramid is between them.  (Should had answered with Cone)
Human: what is next to a cube (should have answered with the likes of cone and pyramid)
Mitsuku: The cone. 

Pretty good. But not quite there. i guess its not a system question but a question of adding more concepts.

But definitely far behind say the 50 year old, Shrdlu.

The problem ofcourse is, just like any other problem. it can’t be scaled up.

How did you picture the 3 objects? The way you have described them could have been:

cube pyramid cone

or

cube cone pyramid

Hence the bot’s confusion. In the first question, the bot appears to have assumed the top position and answered generically, as it couldn’t see any object there. The second question has triggered the bottom position and the bot says that the cone is next to the cube.

Human: what is between a cube and a pyramid
Mitsuku: That which is confined by a cube and a pyramid is between them.  (Should had answered with Cone)

If your answer to the above question is cone, you are describing the objects as cube, cone, pyramid. In which case, Mitsuku’s answer was correct for your 2nd question.

Human: what is next to a cube (should have answered with the likes of cone and pyramid)
Mitsuku: The cone. 

In either situation, the cube can not be the centre object.

i guess its not a system question but a question of adding more concepts.

Indeed but it is also important that the information given to the bot is correct before questioning it. GIGO

Genesis - Aug 7, 2012:

But definitely far behind say the 50 year old, Shrdlu.

I wasn’t aiming to create a SHRDLU type bot. In a similar fashion, Mitsuku is also definitely far behind other programs coded for very specific functions.

Let’s not forget that SHRDLU was designed specifically for these sort of questions about spatial awarness and 3D worlds where my bot is a general purpose one and does this as an extra. Can SHRDLU write poetry, carry out a conversation, compare the attributes of different objects or work out whether 2 plus 3 is the same as 4 plus 1 for example?

Had I concentrated solely on spatial awareness, sure I could have a match for SHRDLU but my bot very rarely gets asked these sort of questions and so I have spent my time ehancing the features people use more often.