Cognitive Code shows off SILVIA artificial intelligence platform

Cognitive Code looks to be taking full advantage of the TechCrunch40 conference to show off its SILVIA artificial intelligence platform, which it calls a "fundamental conceptual breakthrough in artificial intelligence." Apparently, the platform allows for the "development and deployment of intelligent applications to almost any platform" (which we can only hope includes robots), allowing for human interaction in "completely natural and intuitive ways." What's more, it looks like the system should be relatively easy to implement (at least in AI terms), with a complete set of GUI tools available for "developing intelligent entities," and an array of scripting APIs at developers' disposal to embed SILVIA in their applications. If that's not enough, Cognitive Code is also apparently hard at work on a portable version of SILVIA for PDAs and smartphones, with it apparently set for release in late 2008.

[Anwa123 @ Sep 17th 2007 4:48PM]

I love how they spell ihz, i mean is :)

[rrandrews @ Sep 17th 2007 4:48PM]

We'd be reading a snarky comment about this product "fundamental conceptual breakthrough" if the screen wasn't from OS X.

[surfwax95 @ Sep 17th 2007 5:10PM]

That's not OS X, notice the widgets' position on the top-right, rather than top-left. I'm pretty sure that's a custom UI on Windows.

[tdmain @ Sep 17th 2007 5:08PM]

It's not from OS X, the window controls are on the wrong side. It appears to be Windows with some skin thrown on it.

[ryantrevisol @ Sep 17th 2007 5:31PM]

Actually it looks a heck of a lot like Ubuntu or some other Gnome Linux with the Tiger theme.

I'm glad, because it's good to see I've already adopted our future overlord's OS of choice!

[IrishGandalf @ Sep 17th 2007 4:49PM]

"hello dave, hello hal" anyone else see this?

[Jlivengo @ Sep 17th 2007 4:56PM]

She's kinda cute in a 2-d sort of way.

[wakka_wakka @ Sep 17th 2007 5:07PM]

im sorry, gandalf, i cant see that

[Matt @ Sep 17th 2007 11:02PM]

HAL works. The first thing I thought was Max Headroom.

[Xzavier @ Sep 17th 2007 4:59PM]

I love the line where it says... "with a technological core that allows humans to interact with computers in completely natural and intuitive ways"

You mean something besids point and click?

All I want to know is this... Can I turn it off while I am surfing porn? The last thing I want to see is a Javascript taking over this thing!

:D

[Trent @ Sep 17th 2007 4:59PM]

but does she put out?

[Shane @ Sep 17th 2007 5:01PM]

I'm a bit creeped out by that face. Lets hope this AI never develops self awareness...

[MEAT! @ Sep 17th 2007 6:20PM]

You don't have to worry. Any entity that doesn't capitalize her "i"'s isn't self aware. (Just look at texting middle schoolers.)

[ET @ Sep 17th 2007 5:04PM]

Looks like a person w/ down syndrome, is that the level of it's AI?

[Todd @ Sep 17th 2007 5:09PM]

If this is such a break through, why not put it up against the Turing test?

I imagine if it can't spell "is" correctly, it might have a hard time passing...or maybe with today's educations it will help its chances at passing.

[Boynamedsue @ Sep 17th 2007 5:34PM]

the cylons were created by man...

[Ron Smith @ Sep 17th 2007 6:59PM]

you obviously have not seen the spoilers for season four......

[TIMMAH! @ Sep 17th 2007 5:43PM]

The kinder, gentler face of Skynet...

[wakka_wakka @ Sep 17th 2007 5:45PM]

dude, thats not OSX. thats Windows or Linux skinned to look like OSX. look at where the close, minimize, and maximize buttons are, and the font on the buttons at the bottom of the window, as well as the header font and the little application icon in the top left corner of the bar.

i had to point that out.

snark on!

[blackfeather @ Sep 17th 2007 5:45PM]

I wonder how it compares to Jabberwacky:
http://www.jabberwacky.com

Does anyone know if they have an online demo so we can try it out?

[classic white @ Sep 17th 2007 6:59PM]

i hope to god it is better than that. almost 90% of the stuff it says doesnt make sense

[Brian @ Sep 17th 2007 6:30PM]

It'll never eclipse good ol' fashioned ELIZA!!

http://en.wikipedia.org/wiki/ELIZA

[classic white @ Sep 17th 2007 6:30PM]

old news

[anderesp @ Sep 17th 2007 6:35PM]

it reminds me of those really old AIM buddys that do sports, news and really annoying chat responses.


like SmarterChild.
Except this one, i think i can actually manipulate. :)

[Xee @ Sep 17th 2007 7:18PM]

I wonder if it is any better than these:

http://a-i.com/

[joran @ Sep 17th 2007 7:33PM]

it just looks like a glorified chat bot

[Jason @ Sep 17th 2007 7:57PM]

Did this remind anyone else of Dr. Sbaitso? ;-)

[Mimi from CogCode @ Sep 17th 2007 8:35PM]

Thanks to everyone who viewed our SILVIA technology. And yes, it is completely different from chatterbot technology. Instead of finding a corresponding answer to what the user queries, SILVIA responds contextually. See Wikipedia for the problems surrounding chatterbot technology. If you ask a chatterbot something they can't answer, they usually come up with gibberish. Try it, if you will, by going to the ALICE.org website. But, if you ask SILVIA something she doesn't know, she will respond with, "I don't know the answer to that." (In other words, she knows what she doesn't know).

I am still astonished by SILVIA. Sometimes she comes up with an answer that completely takes me by surprise because nobody taught her that. She is still a baby in terms of her knowledge base. Don't forget that the "chatterbots" that have existed have been around for a long long time. Some of them (like Eugene) were the results of a team of AI coders from the top educational institutions in America. And they still haven't nailed down the conversational part yet.

The reason why we had to write 'is' as 'ihz,' is because the voice font was pronouncing it as something else. So we sometimes write it phonetically so that SILVIA will pronounce it correctly.

[jkutsor @ Sep 18th 2007 11:51AM]

Alls you did was add a functionality of "i don't know" response when it in fact doesn't know... I know the ALICE bot can be programmed to do the same or taught to respond that way. I had worked with the chatter bots for 3 years for fun and found that you can make them pretty "fool-proof" relatively speaking. So how does your product stand above the learning functionality of these other A.I.'s? I am interested in this btw...

[DHood @ Sep 18th 2007 1:58PM]

I went to your website (*http://www.cognitivecode.com/news.html) to look for more information concerning SILVIA, but was a little disappointed with the lack of info. Although I did see enough to convince me that SILVIA has the potential to be so much more than the currently availabe chatbots. Is there a video, web interface, demo, etc. that I can interact with to get a clearer picture of what SILVIA is and does. For instance, is SILVIA like the computer interface on Star Trek? Do I have to worry about it going crazy and trying to kill me like HAL 9000? Is SILVIA just a prettied up version of Google Desktop? While I am not convinced that SILVIA is "true" AI (self-aware), SILVIA seems to me to be a much more advanced incarnation of the Office Assistant from M$ Office (please correct me if I am wrong). I think my main problem is that I have no reference point for what SILVIA can do.
Your site says that Desktop SILVIA will be released in Q4 2007, but it hasn't been updated since March. Is this timeframe still viable? What will the cost be? Is it an application for an enterprise (pun not intended, but take out your retainer befor you laugh), or for the average user? I apologize for firing off so many questions a day after you posted, but this was the first opportunity I had.

[Daniel @ Sep 18th 2007 1:55PM]

The "I don't know" answer is a big deal in natural language situations. The reason why is because it takes a much harder algorithm to produce an "I don't know" answer than to produce a statistically significant gibberish answer. Think about what chatterbots do. They generate answers based on keyword and statistical data. An "I don't know" answer is not even possible in the algorithm. Even the most advanced question answering track in the CoNLL doesn't require knowledge of what is in-domain vs. out of domain. The assumption in all the algorithms is that the questions will always be in-domain. At least the most advanced ones are starting to take grammar into account.

[rrandrews @ Sep 19th 2007 6:36PM]

I certainly see the importance or assessing how probable it is that the program's interpretation is correct and being very discriminating with replies. I don't see that as being anything revolutionary, though. You say it's tougher "to produce an 'I don't know' answer than to produce a statistically significant gibberish answer"--but how is your "I don't know" anything other than the program determining that none of interpretations were statistically significant enough to commit to? Sounds like you're just lowering your alpha level to me.

[Shunnabunich @ Sep 18th 2007 12:24PM]

Okay, fine, since nobody else did it:

I, for one, welcome our blue-lipstick'd AI overlords.

[Mimi from CogCode @ Sep 18th 2007 1:05PM]

I thought I posted a reply to the above, but I'm not sure it's showing. Apologies if you see this twice.
To jkutsor - can any chatterbot you know remember your name after an hour? Can you come back the next day after turning off the computer and have the chatterbot remember what you were talking about? Ah...

[blackfeather @ Sep 18th 2007 3:48PM]

Will you let us know when a demo version is available? I'd also love to compare.

[sinnerFA @ Feb 6th 2008 6:06PM]

Actually I have, simple flat files and/or relational databases work wonders....

[Mimi from CogCode @ Sep 19th 2007 1:55AM]

Just to let you know, we gave the MIT Technology Review a copy of the early video we shot of SILVIA. But please remember this, this was shot over a year ago when we were still testing our ideas. She is vastly improved since then. However, in the latter part of the video, you can see her remembering a few items that came up earlier in the conversation.
http://www.technologyreview.com/Biztech/19403/?a=f

[snicolaou @ Dec 13th 2007 8:47AM]

SILVIA seems very promising. Is there is a way we can download a trial version and start testing it?
I have builded a loving personality using an improved version of answerpad. I have spend many months in doing so. I will be happy to see the brain file that I have for answerpad to be used as an input for SILVIA.

[Lord Dolezal @ Sep 30th 2008 10:25AM]

So...what happened to the development of SILVIA? It's been almost a year now and I haven't found anything recent. It looks really promising, so why hasn't there been anything updated or changed since late 2007? Is Cognitive Code still planning on releasing SILVIA?

[Ben @ Nov 6th 2008 9:56AM]

Has Silvia died off? Maybe it was just a big scam?

The website has not been updated in well over a year, come on guys throw us a bone. I'd love to have this controlling my house.