31 replies to this topic

[RighteousReason]

word

[RighteousReason]

wow pretty crazy

[RighteousReason]

Phase II, 2011 – : If viable, design and engineer safe, beneficial AGI, utilizing the theories, tools, and technologies developed in Phase I.

Both of our phases will involve a combination of in-house research and intensive collaboration with researchers in academia and industry.

[Live Forever]

Good stuff.

[basho]

R&D for AGI is really starting to pick up amongst various groups. I feel we at a point where the initial hurdle of doubt ("this is impoosible") has been crossed and people now really do believe AGI is something that can be achieved, and are starting to recognize the incredible first-mover advantage. I expect funding will quietly be increased substantially amongst various government and private organisations.

[maestro949]

I hope I'm wrong in my pessimism but I just don't see a singularity happening in the near-term. What I do see us being able to do is evolve stuff that has intelligence but not understanding how or why because of our own intellectual limitations. I think we'll have to constantly unplug these creations when they go haywire. I don't see it as impossible, I just think the first hurdle is significant extension and enhancement of our own minds' capabilities before we can really understand how to build something smarter than us.

[Bruce Klein]

Ben and I are happy to join SIAI's team!
http://www.singinst.org/aboutus/team

We'll send out the Spring SIAI Newsletter soon w/ more updates...

[RighteousReason]

Ben and I are happy to join SIAI's team!
http://www.singinst.org/aboutus/team

We'll send out the Spring SIAI Newsletter soon w/ more updates...

Hahaha...


wow..

[RighteousReason]

I was under the impression Eliezer wasn't really working with anybody but it seems like Ben G and Novamente are much more integrated into SIAI's ... rather extensive plans ... than it seemed like

[Shannon Vyff]

Great, I signed up for up-dates... thanks for sharing. I loved attending Accelerating Change, I hope to help SIAI in some way in the future...

[Live Forever]

Ben and I are happy to join SIAI's team!
http://www.singinst.org/aboutus/team

We'll send out the Spring SIAI Newsletter soon w/ more updates...

Nice. I am sure you guys will be very beneficial for each other.

[Bruce Klein]

Hoping to see more crossover opportunities to support each other... MF, SIAI, ImmInst, AGIRI, Novamente, FHI, etc...

Btw, two important events are coming up:
http://www.singinst.org/events (Sept 2007, San Francisco)
http://www.agi-08.org (Early 2008, Memphis)

[JohnDoe1234]

Great! a conference somewhat near me! I will most likely be able to attend the one in Memphis...

[RighteousReason]

....YEAH [!]

[RighteousReason]

This shit is crazy

[Live Forever]

Great! a conference somewhat near me! I will most likely be able to attend the one in Memphis...

Ooh, me too. About a 5.5 to 6 hour drive from Atlanta. That isn't too bad. I might try to make it.

This shit is crazy

What "shit" are you referring to, hank?

[JohnDoe1234]

Ooh, me too. About a 5.5 to 6 hour drive from Atlanta. That isn't too bad. I might try to make it.

Yeah, we only live about an hour and a half apart.

[RighteousReason]

well if they can get as much done in 4 years as they are planning to ... shortly thereafter we might be seeing the launch of a Friendly AI

That's the crazy shit I'm referring to!

[JohnDoe1234]

I only hope so... but I still think that it is another 10 years off *at the very minimum*

[DJS]

My hat goes off to SingInst. I must admit that they have far exceeded my expectations of where they would be at circa 2007. Their site looks fabulous!

[DJS]

I think one of the things that needs to be acknowledged is that most of us, myself included, just don't possess the level of technical/theoretical expertise to make even a reasonable guestimate as to time frames and chances of success. For a layperson such as myself, the least one can do is make an effort to become familiar with the basic concepts and terminology. In this regard I am very appreciative of some of the SAI pioneers such as Goertzel and Yudkowsky who have made an effort to dumb down their work, rationale, and motivations to a level where I can understand them.

I am sure that for aspiring technologists the uncertainty of future betting can be quiet disconcerting. To be expected to commit to a career path with nothing more than vague intuitions! This is why I believe that the best advice for newbies seeking career advice is to "do whatever interests you". There are so many technological tracks that could potentially make a difference, and in the end I'm confident that there will be a high level of convergence.

[RighteousReason]

SIAI has apparently laid out something of a timetable (two "phases" of research, 2007-2011, 2011-) based on fairly concrete criteria. They expect to get a huge amount of work done, but I don't think they are necessarily overconfident in their expectations, based on my understanding.

In the 201x's we should see two separate pipelines, one of nanotechnology applications that has already begun now and will continue to accelerate, and then another of AGI applications which mutually accelerate other technology to a point at which RSI-SAI (recursively self-improving super-human artificial intelligence- hopefully a FAI or Friendly AI) is achieved, which then quickly jerks nanotechnology into advanced nanocomputing and advanced molecular machinery and beyond.

... I see all this going down prior to 2020 and later than 2010 , tentatively...

[RighteousReason]

"do whatever interests you".

especially if what interests you happens to be something that can positively influence the shape of the Singularity in a significant way [lol]

[RighteousReason]

In the 2020s, we will be in the computers, in whatever Friendly post-Singularity future [tung]

[DJS]

R&D for AGI is really starting to pick up amongst various groups.  I feel we at a point where the initial hurdle of doubt ("this is impoosible") has been crossed and people now really do believe AGI is something that can be achieved, and are starting to recognize the incredible first-mover advantage.  I expect funding will quietly be increased substantially amongst various government and private organisations.

Yes, Goertzel has been in the field of AI for many years and his account of the prevailing academic opinion on SAI and how it has progressed is informative. Up until a few years ago expressing an interest in AGI was to invite the label of pariah. In large part this could be seen as blow back from the embarrassing failure of GOFAI. However with undeniable advances in our theoretical understanding of cognition (such as connectionsim) and a new generation of researchers that do not have the same bitter taste in their mouths, the close mindedness is gradually starting to dissipate.

The improving credibility and possible feasiblity of AGI is a mixed blessing IMO. Let's not be naive here, the potential military applications of AGI should be obvious to all of us. I'm sure that DARPA already has research projects aimed in this direction. Their specific focus and understanding of AGI may or may not be at the level of independent research teams (certainly their level of funding is superior), but I'm sure they're paying close attention to developments in the field. If novamente or another SAI project gets off the ground do you really think that the powers that be are just going to sit by and congratulate them?

The dynamics for SAI development are very different from those involving anti-aging technologies. Researchers should be cautious and reign in their desire to announce successes. Of course announcing success is very tempting because of the increased funding it could produce, but it may also attract a great deal of... unwanted attention.

[DJS]

especially if what interests you happens to be something that can positively influence the shape of the Singularity in a significant way   [lol]

The fact that our best theoretical minds have differing opinions and areas of focus is very revealing. Aubrey focuses on SENS while Goertzel focuses on AGI.

Why?

Because when one considers highly speculative areas of inquiry, such as technological progress, intuitions will differ. I can recognize an individual as an "authority" in a particular domain of science or technology, but I do not believe there are "authorities" on the extrapolation of technological trends.

Everyone is relying on their intuitions. Of course, having absolute certainty on which technological path will succeed would be lovely, but it's just not the reality of the situation. So what we are left with are differing intuitions. This isn't a totally bad thing. Afterall, it provides diversity of focus and prevents all of our eggs from being placed in one basket.

I am hopeful that AGI will be developed over the next 20-40 years and that it will pay dividends. But if for some reason it doesn't pan out, then having a back up plan (traditional bio tech) that can buy us time is a prudent strategy.

[RighteousReason]

The fact that our best theoretical minds have differing opinions and areas of focus is very revealing. Aubrey focuses on SENS while Goertzel focuses on AGI.

Why?

Because when one considers highly speculative areas of inquiry, such as technological progress, intuitions will differ. I can recognize an individual as an "authority" in a particular domain of science or technology, but I do not believe there are "authorities" on the extrapolation of technological trends.

Everyone is relying on their intuitions. Of course, having absolute certainty on which technological path will succeed would be lovely, but it's just not the reality of the situation. So what we are left with are differing intuitions. This isn't a totally bad thing. Afterall, it provides diversity of focus and prevents all of our eggs from being placed in one basket.

I am hopeful that AGI will be developed over the next 20-40 years and that it will pay dividends. But if for some reason it doesn't pan out, then having a back up plan (traditional bio tech) that can buy us time is a prudent strategy.

Yeah sure. I just think you are underestimating the feasability of AGI.

[RighteousReason]

I also think nanotech is going to accelerate into some really sophisticated systems within the next decade, possibly in the creation of powerful supercomputers and ultimately nanocomputers or really advanced medical applications (cancer cure maybe?)

[RighteousReason]

Once the computers start doing the nanotech research into desinging more powerful computers, or build some kind of nano-seed that assembles the planet into a computronium or something...

[JohnDoe1234]

Hank, you and I have talked and we know each other’s time-frame predictions... but when you say:

I just think you are underestimating the feasability of AGI.

It can easily be countered by: "I just think you are underestimating the complexity of AGI"
Look at the current state of Novamente and A2I2... that is an extremely far cry from a self-improving seed AI... Sure, they further down the line than other AGI systems, but there still is a lot of work to be done, and I think more than what can be accomplished in 4 years... But hey, you can rub it in my face when Starglider/Novamente changes the world 15 years before Sapphire or similar projects reach maturity...