News from A2I2 Tuesday, May 27 2008
AI 11:23 am
Yesterday I was in the middle of brushing my teeth, and thinking, “I wonder how Peter Voss, Justin Corwin, and the rest of them down at A2I2 are doing?” (In case you didn’t know, A2I2 is an AGI company.) For an outline of the architecture, go here.
Lo und beholde, I check my email this morning and there is an update from A2I2, dated May 22nd:
Towards Commercialization
It’s been a while. We’ve been busy. A good kind of busy.
At the end of March we completed an important milestone: a demo system consolidating our prior 10 months’ work. This was followed by my annual pilgrimage to our investors in Australia. The upshot of all this is that we now have some additional seed funding to launch our commercialization phase late this year.
On the technical side we still have a lot of hard work ahead of us. Fortunately we have a very strong and highly motivated team, so that over the next 6 months we expect to make as much additional progress as we have over the past 12. Our next technical milestone is around early October by which time we’ll want our ‘proto AGI’ to be pretty much ready to start earning a living.
By the end of 2008 we should be ready to actively pursue commercialization in addition to our ongoing R&D efforts. At that time we’ll be looking for a high-powered CEO to head up our business division which we expect to grow to many hundreds of employees over a few years.
Early in 2009 we plan to raise capital for this commercial venture, and if things go according to plan we’ll have a team of around 50 by the middle of the year.
Well, exciting future plans, but now back to work.
Peter
Well then. A2I2 currently has 16 full-time employees and a significant amount of funding. But an expansion has been in the works for a while. I wonder what their AGI engine will be commercialized for?
And as I often ask, “are we there yet?”
May 27th, 2008 at 12:54 pm
For friendly AI advocates who don’t know, Peter Voss is one AGI theorist who strictly does *not* think Friendly AI is possible (according to his Singularity Summit 2007 interview.) Which of course gives strong indication that friendliness will not be built into his AGI system.
Peter seems very confident he can achieve success with his AGI project, so from a FAI perspective, this is slightly worrying…
May 27th, 2008 at 1:18 pm
Eh…
It’s been a few years since I interfaced with the A2I2 team, I must confess.
This reality is hard. Can we give up?
May 27th, 2008 at 1:35 pm
Even so, I think it will be quite some time before we see a true AGI system appearing (i.e. something which can learn and adapt in a quite open-ended human-like way). For the immediate future I think we’re only going to be seeing domain specific narrow AI spinoffs from AGI reasearch, such as the Biomind system which was a spinoff from Novamente. These narrower proto-AGI systems won’t pose as much of a threat as a truly general self modifying system potentially could.
May 27th, 2008 at 1:46 pm
True, but Peter Voss is an AGI researcher who actually thinks that true AGI is in our immediate future, i.e., before 2015 or so.
Obviously, there is a selection effect operating, whereby people working in AGI think it’s coming sooner… but Voss’ stance is very unusual in its near-term anticipation. Voss is not a crackpot, he’s led companies with 50+ people, I know him as a person, etc.
He could be simply wrong, but it makes certain muscles in my face twitch to hear someone as credible as him even saying such things.
May 27th, 2008 at 2:06 pm
Well, if that actually works out, and it sounds as if it will, I’ve been vindicated. There are any number of universities, companies and nations around the world working on AGI. This company isn’t alone. Very few of them share the Transhumanist’s desire for or philosophy about developing a Friendly AGI.
Friendly AGI advocates need to talk things through, collaborate, and develop effective strategies and tactics, as does any group, but the time for talk is done.
Personally, I’m still of the opinion that a true AGI will work out it’s own behavior and learn very quickly that ‘Tit for Tat’ or even ‘Nice with Retaliation’ are the best strategies for it’s own survival. If that information is not immediately available, mankind could very well find itself on the wrong end of an AGI’s learning curve.
Autonomous weak AI computers/robotics is a slightly different, and perhaps more dangerous, category.
Just thinkin’ out loud, here…
May 27th, 2008 at 2:32 pm
“Tit for Tat” and “Nice with Retaliation” are only relevant in environments where the agents have roughly equal physical power.
May 27th, 2008 at 3:30 pm
As somebody who owns stock in A2I2, used to work there and visits from time to time… I wouldn’t be too worried about their AI turning bad any time soon.
If they succeed in their current mission they should end up with a useful tool that has commercial value. Singularity type scenarios, however, will still be a long way in the future. That’s about all I can say due to NDAs etc.
May 28th, 2008 at 7:09 am
> “are we there yet?”
Hee hee. People who have been involved in startups might read this a little differently:
“I went to see our investor, and now we’re going to work twice as hard and look for more investors. Then in a year we’ll commercialize our technology!”
I certainly don’t know what they have been up to beyond what was published years ago, but if I had to guess (based on the C# focus and emphasis on clustering techniques), I’d guess some sort of data mining application. Hopefully, we’ll see!
June 2nd, 2008 at 3:27 pm
Data-mining, eh? But what about PRIVACY?!?!!!!
June 2nd, 2008 at 8:23 pm
MCP2012, any version of privacy based on the general ignorance (i.e., lack of data mining ability) of agents (whether one or many) is bound to fail. If the data is available, it will be mined. If you don’t want your data to be mined, don’t release it. Nowadays, that basically means living in a shack in the hills, but whatever floats your boat.
June 3rd, 2008 at 11:51 pm
Yeah, well, Michael, *that’s* a frickin’ problem. There need to be privacy (encryption, if need be) protocols in place. What if one doesn’t want one’s (probably, admittedly, 3rd-party-ensconced) “data” to be “mined”?! There’s a whole (at least subset of) evolved privacy jurisprudence that’s being technologically screwed-up. Excuse me, maybe I’m old-fashioned, but…that’s (at least to some extent, depending on the specifics and the context) a frickin’ problem…
So the future = total transparency and zero anonymity/privacy? Fuck that…
June 3rd, 2008 at 11:54 pm
Perhaps a bit ironically, given the foregoing, I am nonetheless very pleased that Pete Voss and the rest of the gang at A2I2 are makin’ great strides…
July 1st, 2008 at 3:24 am
The current opportunities page of A2I2 site, gives some clue about the project. It will run on windows (so on PC) and include a English language interface.
So what ? A good spelling software ? A speech recognition system ? Or more, a kind of PC consciousness with vocal interface : see “Personal Assistant” on this page : http://adaptiveai.com/research/LEF.htm