Synthetic Intelligence Markup Language

  • 9 Replies
  • 521 Views
*

Zero

  • Trusty Member
  • *********
  • Terminator
  • *
  • 800
  • Ready?
    • Thinkbots are free
Synthetic Intelligence Markup Language
« on: June 11, 2019, 08:19:12 pm »
Hi,

I found SIML (https://simlbot.com/) a few weeks ago. Has anyone tried it? Or have you heard about it, is it good tech, ...etc.? EmotionML sounds sexy!


*

Korrelan

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1418
  • Look into my eyes! WOAH!
    • YouTube
Re: Synthetic Intelligence Markup Language
« Reply #1 on: June 12, 2019, 08:23:41 am »
Hi Zero, nope I've never seen it before, not sure if it looks more suspect than interesting lol.

 :)
It thunk... therefore it is!...    /    Project Page    /    KorrTecx Website

*

Don Patrick

  • Trusty Member
  • ********
  • Replicant
  • *
  • 500
    • Artificial Detective
Re: Synthetic Intelligence Markup Language
« Reply #2 on: June 12, 2019, 09:13:10 am »
It is a derivative of the AIML chatbot scripting language with more javascript integration, released some years ago. It's not interesting in terms of AI. The "ML" stands for "Markup Language" rather than machine learning.
https://developer.syn.co.in/tutorial/siml/siml-aiml.html

The term "emotion" appears to refer to variables like "happy" that you can set to a value to take into account in the rest of the script. Consider me underwhelmed.
https://developer.syn.co.in/tutorial/bot/siml/emotions.html
Personal project: NLP -> learning -> knowledge -> logical inference -> A.I.

*

Zero

  • Trusty Member
  • *********
  • Terminator
  • *
  • 800
  • Ready?
    • Thinkbots are free
Re: Synthetic Intelligence Markup Language
« Reply #3 on: June 12, 2019, 09:45:15 am »
hi Korrelan, hi Don Patrick,

I guess you mean "not interesting in terms of genuine AI"?

Apparently, SIML lets you code intelligent behaviors, and has (probably limited) learning capabilities.

*

Don Patrick

  • Trusty Member
  • ********
  • Replicant
  • *
  • 500
    • Artificial Detective
Re: Synthetic Intelligence Markup Language
« Reply #4 on: June 12, 2019, 11:28:44 am »
Hi Zero. As a rule I don't hide behind words like "genuine" or "true". My bar for calling things AI is already lower than that of most people, but I don't consider memorising in that manner learning. It is explicitly instructed which words to remember and exactly what to do with them. If I considered that learning, then I would also have to call an Excel sheet intelligent because it can store a value from one cell and display it in another cell that's rigged to display it. By my definition, learning is not the product of an explicit instruction, but from a broader analysis.

Of course you are free to find it interesting yourself, but then you should find AIML just as interesting because it can do these same things.
Personal project: NLP -> learning -> knowledge -> logical inference -> A.I.

*

Zero

  • Trusty Member
  • *********
  • Terminator
  • *
  • 800
  • Ready?
    • Thinkbots are free
Re: Synthetic Intelligence Markup Language
« Reply #5 on: June 12, 2019, 01:09:58 pm »
I agree on AIML being "just as interesting", indeed they are similar tools.

There are, broadly speaking, 2 types of learning. The first, primitive way of learning, is very low level: reinforcement learning, pattern matching, Pavlov, ...etc. The other way is symbolic learning: by means of words, explanations, deductions, ...etc. I tend to think human beings gain their high level knowledge mainly through symbolic learning.

Now, there's a huge gap between SIML, which is, as you said, explicitly instructed what to "learn", and humans of course. Symbolic learning is certainly not about stuffing variables with values! Still I believe Symbolic AI's not dead. Like punk.

*

Don Patrick

  • Trusty Member
  • ********
  • Replicant
  • *
  • 500
    • Artificial Detective
Re: Synthetic Intelligence Markup Language
« Reply #6 on: June 12, 2019, 07:03:49 pm »
My own project is also "symbolic" AI, but all that term really means is that it happens to use words as its carrier for information. IBM's Watson was mainly symbolic AI, but it was a far stretch from pattern matching. The lines do blur at some point, as AIML's Mitsuku can perform several inferences and uses a small knowledge database, but I don't consider pattern matching systems to be suitable tools for that job.
Personal project: NLP -> learning -> knowledge -> logical inference -> A.I.

*

goaty

  • Starship Trooper
  • *******
  • 316
Re: Synthetic Intelligence Markup Language
« Reply #7 on: June 12, 2019, 09:55:48 pm »
I have to look at all your projects now,  cause I just accidentally opened up a can of worms about symbolic ai myself.

Don Patrick, I love your attention to detail, and I know that ur system doesn't respond logically sometimes, but uve definitely got some good symbolic stuff there!  I bet if you stay in its realm of understanding it goes perfect,  but im yet to try it again being a bit more friendly to it.

If this "unit" has a predetermined response to carry out how its doing *ANYTHING* which includes what to remember, how to parse it, my order of delivery, to me this is still classed as goal based behaviour,  the only difference is it didn't have to develop it,  it was granted the ability,   and robots such as these can be just as amazing as ones that don't. 

I think that if its just "programmed" it can still be artificial general intelligence!  because with a good system around it, to help it see where it currently is, a program can just take over and it can do any task.  that's general isn't it?  I know that its more predictable, and its boring,  but it can carry out all routines, they just have to be given to it.

But if you want to say,  this is not a.i.,  to me,  you may convince me sometime in the near future, 
because the catch is,  ------it has to work it out, without you typing it directly, to be a.i.,------- 

not just a fancy engine system that can see and is just running symbolic programs converting to direct motor.  which isn't a.i.  ;)



*

Zero

  • Trusty Member
  • *********
  • Terminator
  • *
  • 800
  • Ready?
    • Thinkbots are free
Re: Synthetic Intelligence Markup Language
« Reply #8 on: June 13, 2019, 09:28:27 am »
Hi,

The way I see it, the first steps can be made through "similarity learning", which produces behaviors & symbolic knowledge. Once symbolic knowledge has reached a critical mass, it can produce learning rules by itself. Then it goes logic-style.

In fact the system should learn how to learn.

You have a project of your own Goaty?

*

Don Patrick

  • Trusty Member
  • ********
  • Replicant
  • *
  • 500
    • Artificial Detective
Re: Synthetic Intelligence Markup Language
« Reply #9 on: June 13, 2019, 11:29:16 am »
When my program doesn't respond logically it is almost always because the language processing misinterpreted something, rather than the logic systems.

it has to work it out, without you typing it directly, to be a.i.
That's more or less my point, yes. The degree of intelligence depends on how specific the programmed instructions are. AI is necessary because it's impossible to manually specify every detail of every solution to every problem that doesn't exist yet.
Personal project: NLP -> learning -> knowledge -> logical inference -> A.I.

 


Users Online

47 Guests, 0 Users

Most Online Today: 74. Most Online Ever: 340 (March 26, 2019, 09:47:57 pm)

Articles