Hi everyone,

I’m from a startup called Snasci.  This may come as a bit of a shock, but we’re launching a real Artificial General Intelligence.  Its not a chatbot, its not a personal assistant, it is a full on AGI just like you see in the movies.  Reasoning, conversation, planning, etc.

We have just come out of stealth mode after nearly a decade in development and we’re launching an Indiegogo campaign to move our software into the cloud.  We’ve scheduled the campaign to begin in February and we’re in the process of a buildup towards that.

We have a lot going on, from social media to providing background information on Wordpress.  So excuse us if we’re a bit pressed for time.  That said, hit Google and start looking for the links, I’ll try to answer as much as I can as I get time.

No proof- lol!

“...our current designs ... used to disable any Snasci device completely.  We expect this to fail under two independent conditions, the first is a coded radio signal (recoverable) and the second is electrical overload (unrecoverable). Authorised law enforcement will thus have the capability to disable any Snasci-enabled device.  These designs will be open and be a requirement for any manufacturer to implement before approval for Snasci integration will be provided.”

-and anyone else who can hack your device/robot!

This sounds fishy at best.

With all due respect, I have my doubts that a ‘real’ AGI has been developed without having been leaked even if in stealth mode. The NSA works and remains in stealth mode and look what happened to it.

A real AGI would be ground breaking and every major player in the world that mattered would want to get their hands involved in some manner or another. It should also be a working entity, not some scribbled notes on paper that we often find.

Please excuse the skepticism as we have all heard it before. The latest, greatest thing since magnets or whatever.

I really hope you Have a working prototype of a real AGI.  If you are unable to provide some kind of proof, then you and your “team” are likely to be labeled, Fortune Hunters, like so many others have been on those “crowd funding adventures”.

Good luck to you.

I completely appreciate the healthy scepticism and you are completely correct to take that position.  Nonetheless, I’m making it clear that it has been achieved and the migration begins in February.  There will be a demo video of a prototype of Snasci before then conversing in text mode.  I’m finishing the code for this proof-of-concept as we speak.  This will be the first time that I’ve stitched the entire thing together, but all the individual components are working.  No doubt there will be some bugs and edge cases, but generally speaking every use case I’ve have thrown at the design has had no problems.  I have sat down and examined numerous reasoning scenarios, from deriving the context of a conversation right through to inference of implicit information (including over time) and Snasci’s architecture has no problems with it.  I can clearly see in code where to obtain this information, stitch it together and perform actions based upon it.

From Snasci’s perspective, the main challenge will be speed, not cognitive capability.

Let me clarify this thing about stealth mode.  It just means it was hidden as a commercial project for a long time.  The underlying grid computing platform’s R&D and composable nature was done openly at a university and I have had brief discussions with OpenAI and Y-Combinator on this matter earlier this year.  So, Snasci hasn’t been entirely dark for the last several years.  Obviously, I keep cards close to my chest, but that’s just a business thing. 

I did, however, have a recent breakthrough in the representation of data which dramatically reduces the training time, whilst opening the platform to a collaborative paradigm (i.e. Snoodles).  This was really an evolution of something I was working on the last number of years.  So, that may have caught people by surprise, as they are probably still thinking of AGI in terms of Google’s DeepMind and biologically inspired designs which are extremely complex and costly to create.

All I can really say at this point is, get ready for this…and a world full of annoying talking toys, melodramatic phones, and impudent domestic appliances.

Thank you for your positional explanation. For me it’s always better to err on the side of caution. If I was premature or unwarranted in my skepticism, I apologize.  We have seen far too many inflated claims in the past that deflate in a few months.

Mr. Leslie Spring of Cognitive Code developed a very clever and quite useful system that enabled his program to be able to converse rather fluidly and instinctively on any given topic. (I think the key word here might be ‘given’, but I could be mistaken). He started several years ago on a laptop with a demo conversation of his system, Silvia. Now he’s involved with governments and corporations.

You obviously know that any program that has sufficient abilities and promise such as your that you’ve described will be of great value to the Government and Military, in that order. If that’s your plan, fine, but there are also many humanitarian groups that could benefit from such a system. I’m sure you’ve thought of many possibilities.

Best of luck with it. Can’t wait to see the demo at some point.

- Thanks! -

No doubt every scammer tries to have a go.  I’ve seen a lot of technologies over the years pitched as the next big things, only for it to be not well thought through as a business model, some form of scam, or merely a naive implementation.  Scepticism is good, because it hits people like a bomb when you deliver.  That singular jaw dropping moment.  That’s special.

Snasci will have have absolutely no role with military or policing of any form, ever.  It may have a role with broader government, but in an entirely civilian capacity.  You can check out our articles on Wordpress which details why this is.  In short though, governments will be wary of other governments and it would lead to all sorts of complications.  Best just to stay out of it.

Hopefully, if all goes well, the first basic demos should be done by sometime in early January.  Right now, there is something like 26 different projects that need to be wired together to a full Snasci instance, but I think we can get away with much less to demonstrate core competencies.  For example, do I need to demonstrate humour and emotion at this point?  Aren’t they just really really modifiers to core competencies? How about open-ended discussion?  Isn’t that just a flow of core competencies?  So, if I focus on just demonstrating comprehension, knowledge retrieval, some inferences, etc., it should prove the point that the rest of it works, as they are all built on top of this layer.

That should be easy enough to deliver at this point.

I’ll have a look when there’s something to see. I’m sure you must have built something in a decade, but I’m also sure everyone’s using their own personal definition of AGI. It’s a problematic thing to claim and meaningless in practice, as only its abilities will matter.
Lots of AI projects do logical inference (e.g. Cyc), so if you want to stand out as “AGI”, you’d definitely have to show more than the basic building blocks. The common public will be more easily convinced if indeed you showed its abilities with open-ended conversation, emotion and humour, even if technically they are just more of the same.

Art I almost got excited when I read your comments about SILVIA and went and researched it. Sadly it just seems to be yet another AIML/Rivescript/ChatScript scripting system, albeit well packaged and integrated with the Unity graphics engine. Founded eight years ago, if they had anything significant to offer I think they would have taken off in a big way by now. But, they haven’t, it’s just another really fancy website with a misspelled word on it.

I only wish all new members on these A.I. forums would just lurk for a few months and see what we have to contend with before jumping in with their outrageous claims. If they all saw how much rubbish gets posted here every week they’d think twice before wasting everyone’s time, and, they’d be in a better position to differentiate themselves so we might actually take them seriously.

Snasci to your credit you did not start out by disparaging the rest of the industry and everyone in it. That’s the fatal mistake that gives away most crackpot inventors in their first post. Your claims are still a bit hard to accept though. There’s a vast difference between the toy systems that prove a concept in the lab, and industrial scale systems that are necessary for commercial success. The original Siri was orders of magnitude smarter than the software that made it onto iPhones after it was stripped and gutted to make it workable on such a large scale.

Don Patrick - Dec 3, 2016:

I’ll have a look when there’s something to see. I’m sure you must have built something in a decade, but I’m also sure everyone’s using their own personal definition of AGI. It’s a problematic thing to claim and meaningless in practice, as only its abilities will matter.
Lots of AI projects do logical inference (e.g. Cyc), so if you want to stand out as “AGI”, you’d definitely have to show more than the basic building blocks. The common public will be more easily convinced if indeed you showed its abilities with open-ended conversation, emotion and humour, even if technically they are just more of the same.

To get a better idea of Snasci, check out our Wordpress site:

https://snasci.wordpress.com/

Read the current article on why we don’t use deep learning.  It will begin to break down some of the preconceptions people have about AGI.

I don’t think the latter 3 are really the definition of an AGI.  Ultimately, emotion and humour are just tricks of classification and word/image/video responses.  A narrow AI can do that.  To me, an AGI is one that describe any problem in a universal format and process it.  Further, that it can use this format itself to advance its own knowledge and capability.  Some use deep learning for this, I took a different approach.

This is something I just polished off yesterday in the prototype code.  So, I’ll demo Snasci going through some problems and show it working things out.

Andrew Smith - Dec 3, 2016:

Snasci to your credit you did not start out by disparaging the rest of the industry and everyone in it. That’s the fatal mistake that gives away most crackpot inventors in their first post. Your claims are still a bit hard to accept though. There’s a vast difference between the toy systems that prove a concept in the lab, and industrial scale systems that are necessary for commercial success. The original Siri was orders of magnitude smarter than the software that made it onto iPhones after it was stripped and gutted to make it workable on such a large scale.

My professional background is developing scalable cloud solutions for enterprises, so I have the skillset to architect this at scale.  Speed is something I have been concerned about from day one and whilst it is not required for the proof-of-concept code, I’ve have numerous robust architectures for when it is migrated into the cloud.  Ultimately, even then there is the option to express most of it in hardware and drastically cut latency and processing times.  Once demonstrated that it works, obtaining VC funding to produce such hardware should be fairly easy.

Right now, I don’t think hardware will be required.

Most of the prototype is now done and a quick teaser trailer of Snasci talking around Christmas is expected at this point. A simple hello world, but under the hood the brain will be working, so this won’t be a mock up of any form.

If you have time, head over to our Wordpress site.  There are a number of articles now exploring various topics that we’ve had to consider in this design process:

https://snasci.wordpress.com/2016/11/28/welcome-to-snasci/
https://snasci.wordpress.com/2016/11/30/a-deep-dive-into-the-high-level-architecture-of-snasci/
https://snasci.wordpress.com/2016/12/01/agi-ethics/
https://snasci.wordpress.com/2016/12/05/data-layer-architecture-security/
https://snasci.wordpress.com/2016/12/07/why-does-snasci-not-use-deep-learning/
https://snasci.wordpress.com/2016/12/11/classification-as-a-service-what-it-means-for-your-business/
https://snasci.wordpress.com/2016/12/15/honour-bound-ethically-reactive-security-model/

If you’re still wondering how Snasci works, well I’ve provided a tiny bit of insight into this.  Over at Twitter, I have explained that we’ve invented a 6th generation programming language called Deep Intelligence and that our product Snoodles is the IDE/debugger for this.

https://twitter.com/SnasciPlatform

If you want to sign up to our Indiegogo project for updates, here are links you can use:

http://www.pitchfuse.com/project/snasci-artificial-general-intelligence/1239
http://prefundia.com/projects/view/snasci-artificial-general-intelligence/11421/
https://www.indiegogo.com/projects/snasci-artificial-general-intelligence-software/coming_soon

If you like to become part of the Snasci Beta Tester team, then you can signup at either of these locations:
https://www.betafy.co/project/view/snasci-artificial-general-intelligence
https://snasci.com/AI/TesterSignup

If you want to catch up with us on social media, use these links:
https://www.youtube.com/channel/UCZ05V6BwaMe1JUDFARWDmkg
https://plus.google.com/+Snasci
https://twitter.com/SnasciPlatform
https://www.facebook.com/snasciai
https://snasci.wordpress.com

Snasci - Dec 21, 2016:

Most of the prototype is now done and a quick teaser trailer of Snasci talking around Christmas is expected at this point. A simple hello world, but under the hood the brain will be working, so this won’t be a mock up of any form.

This sounds laughable. What do you expect it to prove?

I’m thinking “Baby Steps” here, but that might just be me.

Dave Morton - Dec 21, 2016:

I’m thinking “Baby Steps” here, but that might just be me.

If they were beginners, or if they had done or said a single thing that established their credibility, I’d be full of encouragement. However they are claiming to be major players, producing prodigious amounts of meaningless marketing material (I’ve been through it all and it’s all garbage) and preparing to ask for lots of money from lots of people. It has scam written all over it.

Snasci is a pipe dream at the moment.  The code base for “SnasciVoice” on Github is just a template. There is no interface or demo for the “AGI”.

*sigh*

Tentative Release date: 01Apr2017

On Dec. 2, he stated: “Hopefully, if all goes well, the first basic demos should be done by sometime in early January.”
===========
The fact that only a “template” of a SnasciVoice exists is certainly no proof that this creation does or does not exist.

Tolerance is based on having an open mind toward any concept no matter how far removed from our own beliefs it might seem to be.

Though we have seem many come and go, many claiming greatness like some of the prize fighters of the past proclaimed many years ago, to those who were only a flash-in-the-pan…a legend in their own minds.

I for one, will put aside any preconceived notion of doubt until I am swayed otherwise by any forthcoming evidence, or lack thereof.

Time, as they say, will tell.