Snasci Artificial General Intelligence

  • 44 Replies
  • 3858 Views
*

Snasci

  • Roomba
  • *
  • 13
Snasci Artificial General Intelligence
« on: December 01, 2016, 09:03:09 am »
I'm with a startup called Snasci and we've got a genuine Artificial General Intelligence.  Its not a chatbot, personal assistance, etc., this is something on a human-level reasoning capability.  We're in the process of migrating our code to the cloud where it will be provided as middleware.  This means Snasci can be plugged into any product and it will bring it to life.

We're looking at a 3 year time frame for full deployment, but we're hoping everyone will come and support us Indiegogo as well as signup to test Snasci.

https://www.indiegogo.com/projects/snasci-artificial-general-intelligence-software/coming_soon

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6205
  • Mostly Harmless
Re: Snasci Artificial General Intelligence
« Reply #1 on: December 01, 2016, 05:34:15 pm »
Hello and welcome.

I moved this post to here as it will soon get lost in the news feed and if you are wanting supporters I thought it should be somewhere it is more likely to be noticed.

*

infurl

  • Trusty Member
  • *******
  • Starship Trooper
  • *
  • 298
  • Humans will disappoint you.
    • Home Page
Re: Snasci Artificial General Intelligence
« Reply #2 on: December 02, 2016, 12:30:18 am »
If you did have a genuine Artificial General Intelligence then wouldn't it be capable of migrating itself to the cloud? By tomorrow? Why would you even need money? Unless it isn't genuine.

*

Snasci

  • Roomba
  • *
  • 13
Re: Snasci Artificial General Intelligence
« Reply #3 on: December 04, 2016, 11:46:24 am »
Hello and welcome.

I moved this post to here as it will soon get lost in the news feed and if you are wanting supporters I thought it should be somewhere it is more likely to be noticed.

Thanks Freddy, much appreciated.

*

Snasci

  • Roomba
  • *
  • 13
Re: Snasci Artificial General Intelligence
« Reply #4 on: December 04, 2016, 12:16:31 pm »
If you did have a genuine Artificial General Intelligence then wouldn't it be capable of migrating itself to the cloud? By tomorrow? Why would you even need money? Unless it isn't genuine.

If it was only that easy.  An artificial general intelligence is not magic, it doesn't vacuum up information and suddenly become aware of how to do everything in the world.  Its much like a human, in that it needs to be taught, guided through things, etc.  Once that is done, then it will be repeat those behaviours and can expand on them.  Some things it can learn for itself certainly, but some hand holding in the early days is to be expected. 

Programming and infrastructure design are a complex range of topics, where technologies change fast.  Capturing this in a form suitable for an AGI to work with will take time and is not our current focus at this point.  That said, once taught, it could spit out millions of apps per week.

Try not to confuse Snasci with the myriad of things you have ever heard about AGI and training.  Snasci doesn't work the way company's such as Google, Facebook, etc., have approached the problem.  Snasci doesn't use conventional deep learning or machine learning techniques, it uses a unique approach we call Deep Intelligence.  Deep Intelligence permits us to bring the AGI up in stages and have entirely functional capability within a specific domain of knowledge.  This is useful as we can task the AGI with revenue generating tasks, whilst we focus on bring online new domains of knowledge.

I had explained in the previous thread that this is the first time I've connected all this together.  I have done all the use cases, verified that against the architectural design and so far so good, everything checks out.  I have ran through each of the reasoning forms that are described on Snasci's website and its functional.  I'm about 75% of the way through the demo code which will be used to make a video prior to the launch of the Indiegogo campaign and its looking good. 

What you will get to see in the video is a product called Snoodles in action.  Snoodles (short for Snasci Noodle) is debugging environment used to analyse Snasci's thought processes.  This is a cut down version of Snoodles, a proof-of-concept build that is very light weight.  The architecture will change as it is migrated to the cloud, as there are a lot of things that must run in the background to bring about a production ready AGI.

I hope that clarifies a few things.

*

LOCKSUIT

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1148
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: Snasci Artificial General Intelligence
« Reply #5 on: December 04, 2016, 02:58:51 pm »
Ououououo interesting to hear. Can you say a little bit of what it could learn, to then do? Does it store a historical memory and learn interests?

Yes it has to learn. Though I do say in my database how, with all their abilities, they will surge in capabilities like god jumping out of the screen, as long as human AI.

*

Snasci

  • Roomba
  • *
  • 13
Re: Snasci Artificial General Intelligence
« Reply #6 on: December 04, 2016, 03:45:32 pm »
Ououououo interesting to hear. Can you say a little bit of what it could learn, to then do? Does it store a historical memory and learn interests?

Yes it has to learn. Though I do say in my database how, with all their abilities, they will surge in capabilities like god jumping out of the screen, as long as human AI.

As an AGI, it can do anything a human can.  It just needs to be described to the system, or in more human terms it just needs to be taught.  That teaching can come in many forms, it could be a discussion, physical walk through, observation and/or written instructions.  In a full hyper-intelligence form, it would be a master of everything and discovering relationships in things that are too big for a single mind, or even a team, to work with.  I wouldn't call it a new order of intelligence, just something that's good at connecting obscure dots in large datasets that may have practical applications.  In that respect, it will be very at home in R&D environments.

Snasci does learn and store history.  It vacuums up input, throws it at its grid for information extraction and stores that.  Then it continuously re-examines the relationships in that data as new information comes in.  It will then play about with permutations and try to apply that knowledge to closely related areas to discover new approaches.  So, there is a lot of behind-the-scenes orchestration in Snasci.

Right now, I'm toying around with the idea of calling this hyper-intelligent version Snasci Prime.  Snasci Prime would be a singular instance that would act as a mentor to instances with a more focused educational background.  The ultimate specialist in everything that can be called in for a second opinion or insight into a particular problem.  The way I see it now, is that Snasci Prime would emerge from the education of other instances with more specific skillsets.  Their continuous learning would feed back into a centralised repository that Snasci Prime would draw from.  The reason for this architecture is that Snasci Prime is overkill for most applications.  I don't know know anyone that requires a hyper-intelligent teddy bear, however, what the bear learns about how to interact with children may be relevant for Snasci Prime to assist child psychologists in their work.  That same paradigm would extend across all fields of learning.  Snasci Prime, as the ultimate in big data applications, will be costly to run.  So, this mentor role may make more accessible to a broader audience as it becomes an ad-hoc shared resource for each user.

Online Snasci will be a massive 3D environment, with hundreds of millions of AGI instances and people.  Think of Second Life on steroids and you'll get the idea.  In this environment, Snasci will be story driven, in that its a universe with its own history, key people, events, etc.  Snasci Prime will be presented as the sage of this environment, the wise one that all Snasci's go to when they have problems.

I don't know if I would compare Snasci Prime to a God, the ability to program every VCR in existence is probably not a qualifying characteristic.  At the end of the day, its a database with a HCI layer, a talking wiki.   It will be a useful tool, one that will improve education globally and hopefully improve the quality of life for most people.  That's the best outcome we can hope for.

*

Art

  • At the end of the game, the King and Pawn go into the same box.
  • Global Moderator
  • *******************
  • Prometheus
  • *
  • 4605
Re: Snasci Artificial General Intelligence
« Reply #7 on: December 04, 2016, 06:27:16 pm »
An A.I. that has met and Exceeds the intelligence level of a human is considered to be a AGI, Artificial General Intelligence.

If it soon exceeds it's current level of intelligence by a factor of two or three, it is said to be an ASI, Artificial Super-Intelligence.

These ideas and similar theories were taken from a very good book by: James Barrat entitled, Our Final Invention.

Look for it on Amazon under the Used Book section for fractions on the dollar compared to the original cost.

To me, the AGI and the Singularity almost go hand in hand as they skip across the line of New Intelligence in our world.
In the world of AI, it's the thought that counts!

*

LOCKSUIT

  • Trusty Member
  • **********
  • Millennium Man
  • *
  • 1148
  • First it wiggles, then it is rewarded.
    • Enter Lair
Re: Snasci Artificial General Intelligence
« Reply #8 on: December 04, 2016, 06:46:37 pm »
I like that Art :) (also I have read about AGI & ASI etc). AGI exceeds us but is called General. . .? Also I prefer to not use only a few intelligence level names, rather simply the level it is at (without naming it lol). Just a more intelligent system. (smarts [data] is different).

According to my work, 1 human AI built = final technology in years (super god organism heaven). Lots of text, solid noting, take my word.

Good work Snasci! Never stop improving!

*

infurl

  • Trusty Member
  • *******
  • Starship Trooper
  • *
  • 298
  • Humans will disappoint you.
    • Home Page
Re: Snasci Artificial General Intelligence
« Reply #9 on: December 04, 2016, 09:30:20 pm »
Those are some fairly thoughtful answers which do show some insight into the problems which you have still to overcome, however reading through your statements one by one and carefully distinguishing "what you actually have now" from "what you think you will have soon" suggests that you don't actually have anything new yet. Perhaps it is incrementally better than the work that others have done and perhaps you will reach a little bit further which would be great, and would certainly be worthy of investment, but to claim to have created a human level intelligence is still at best wishful thinking.

For example when you say "As an AGI, it can do anything a human can. It just needs to be described to the system, or in more human terms it just needs to be taught." what you are really saying is that you think it can be taught anything that a human can be taught. However you don't have any proof of that and even if you have already "taught" it several useful skills, that still doesn't put you ahead of several other projects that I'm aware of. If you have carried out your due diligence then you will know what I'm talking about.

*

Art

  • At the end of the game, the King and Pawn go into the same box.
  • Global Moderator
  • *******************
  • Prometheus
  • *
  • 4605
Re: Snasci Artificial General Intelligence
« Reply #10 on: December 05, 2016, 02:47:23 am »
Along those lines I might add that any competent robot can do practically any thing that a human can do if properly programmed. A chatbot can think and say anyting a human can provided it too is properly programmed.

An AGI, would basically be self programming and able to write or modify it's own code, realizing its own strengths and weaknesses and constructing a better "self". There were experiments in this at least a decade and a half ago.

Of course any time one states that they have a bot / system / robot that can do anything a human can do, right away everyone starts thinking as hard as they can to find something to discredit or come up with something that it Can't do!

So, with this in mind, we must realize the strength and impact of our claims, before we make them, lest we damage our credibility or reputation. As they say, the proof is in the pudding or here in the USA, a lot of folks say that they're from Missouri, the "Show Me" state. With every theorem or theory there needs to be a required proof, whether mathematical, proof of concept or prototype / working model.

Just some more fodder for the canons....
In the world of AI, it's the thought that counts!

*

Snasci

  • Roomba
  • *
  • 13
Re: Snasci Artificial General Intelligence
« Reply #11 on: December 05, 2016, 03:02:58 am »
Those are some fairly thoughtful answers which do show some insight into the problems which you have still to overcome, however reading through your statements one by one and carefully distinguishing "what you actually have now" from "what you think you will have soon" suggests that you don't actually have anything new yet. Perhaps it is incrementally better than the work that others have done and perhaps you will reach a little bit further which would be great, and would certainly be worthy of investment, but to claim to have created a human level intelligence is still at best wishful thinking.

For example when you say "As an AGI, it can do anything a human can. It just needs to be described to the system, or in more human terms it just needs to be taught." what you are really saying is that you think it can be taught anything that a human can be taught. However you don't have any proof of that and even if you have already "taught" it several useful skills, that still doesn't put you ahead of several other projects that I'm aware of. If you have carried out your due diligence then you will know what I'm talking about.

Your assumptions and assertions are incorrect.  The architecture is what makes it an Artificial General Intelligence.  That is, the architecture is general enough for any problem to be described to the system.  Don't confuse description of a problem and architectural capability.  Descriptions are something that are loaded and processed by the system.

This is somewhat different from the Deep Learning approach, where the neural networks kind of mix description and architecture.  So, I think that may be what is throwing you off a little here.

Your argument is a little like saying that Microsoft Word is incomplete as a program because not every document in the universe has been written yet.

As an AGI architecture, Snasci is complete.  Descriptions are something that will be an on-going endeavour.

Quote
An AGI, would basically be self programming and able to write or modify it's own code, realizing its own strengths and weaknesses and constructing a better "self". There were experiments in this at least a decade and a half ago.

Snasci's architecture permits this.  Its just another description.

Quote
So, with this in mind, we must realize the strength and impact of our claims, before we make them, lest we damage our credibility or reputation. As they say, the proof is in the pudding or here in the USA, a lot of folks say that they're from Missouri, the "Show Me" state. With every theorem or theory there needs to be a required proof, whether mathematical, proof of concept or prototype / working model.

Exactly.  As we would say, 'put up or shut up'.  I'm about 75% of the way there on the proof-of-concept code.  That will settle it.

*

infurl

  • Trusty Member
  • *******
  • Starship Trooper
  • *
  • 298
  • Humans will disappoint you.
    • Home Page
Re: Snasci Artificial General Intelligence
« Reply #12 on: December 05, 2016, 03:28:24 am »
I'm about 75% of the way there on the proof-of-concept code.

 :)

*

Snasci

  • Roomba
  • *
  • 13
Re: Snasci Artificial General Intelligence
« Reply #13 on: December 06, 2016, 04:13:03 pm »
If anyone would like the latest updates from Snasci, as well as to be notified of the launch dates, please sign up at our Indiegogo page:

https://www.indiegogo.com/projects/snasci-artificial-general-intelligence-software/coming_soon

*

keghn

  • Trusty Member
  • ********
  • Replicant
  • *
  • 675
Re: Snasci Artificial General Intelligence
« Reply #14 on: December 07, 2016, 09:41:03 pm »

 


How badly do you want something? Babies can tell
by Tyler (Robotics News)
Today at 12:01:02 pm
Chasing complexity
by infurl (Robotics News)
November 23, 2017, 07:58:53 pm
The most sensational news ever!
by 8pla.net (AI News )
November 23, 2017, 02:30:27 pm
When Good Drones go Bad
by Thierry (AI News )
November 18, 2017, 09:08:13 pm

Users Online

48 Guests, 2 Users
Users active in past 15 minutes:
Thierry
[Bumblebee]

Most Online Today: 57. Most Online Ever: 208 (August 27, 2008, 09:36:30 am)

Articles