SecondLife’s CEO Starts a New AGI Company Saturday, Feb 6 2010
bioethics 4:02 pm
From New World Notes, James Au’s standard blog on SL, comes news that Philip Rosedale, founder of Linden Lab (the company behind SecondLife), is “working towards creating a sentient artificial intelligence which [exists] in a virtual world.” Um, OK. Here is a quote:
That Philip plans to revolutionize AI technology — in effect, achieving singularity in a virtual world — isn’t that surprising, because he said as much when I talked with him for The Making of Second Life:
Hey! Slow down. A Singularity in a virtual world could easily transcend its boundaries and cause problems for us all. I don’t think that Rosedale will achieve human-level AGI anytime soon, but we must acknowledge that the boundary between the virtual and the “real” is pretty damn thin. Perhaps this is just James Au’s choice of words, but I wish all of his readers would understand that AGI is not just a game, or a new entertainment company. It could utterly transform the planet in a way that it’s never been transformed in its entire 4.56 billion year history.
February 6th, 2010 at 7:20 pm
This illustrates the fact that one of the main sources of danger comes from the competitive imperatives that capitalism engenders, rather than from the more esoteric possibilities that have been imagined.
Are you not more likely to play fast and loose if you think you can make a quick squillion?
The temptation of economic entities under capitalism to flout the precautionary principle is obvious. But I wonder even whether the proactionary principle is upheld…since this seems logically to require that agents (or, at least many, well-funded ones) are doing their best to achieve good/safe ends, whereas that is not the PRIMARY focus for a value-maximizing entity.
February 7th, 2010 at 5:26 am
Didn’t you consider scenario in which it is… a SCAM? Come on, anyone that presents you vision of AGI’s “behind the corner” is a lunatic or smebody preparing a swindle. Besides, it is not that difficult to emulate brain function’s, aka inteligence. AIXI can do the same-and what, he is so “smart” that he doesn’t understand concept of morality and you cannot teach him that (at least today). Algorithmic inteligence is a marvelous idea, but it will just give you dangerous, dumb as brick geniuses. One mistake in programming and “ooops, I think we have SkyNet”. Real inteligence is able to reach correct conclusions despite of, or being based on wrong asumptions. And what I read in this news is “we will have really smart bots in Second Life, isn’t that cool?”. Yes, its cool. But nowhere near singularity.
February 7th, 2010 at 10:12 am
It could utterly transform the planet in a way that it’s never been transformed in its entire 4.56 billion year history.
Well, a certain specie achieved general intelligence some time ago and transformed the planet in ways that are starting to accumulate and explode. This specie is now beginning to develop AGI, which will transform the planet in even more radical ways.
smart bots in Second Life…
Then smarter bots, then very smart bots, then even smarter bots, then reeeeally smart bots… I have always been sort of persuaded that real AGI may well emerge from the entertainment market.
February 7th, 2010 at 12:24 pm
An obvious but still clever way to develop and test socialized AGI: inside of a “VM” world which, considering Second Life, might conveniently have humans around with which to train and (especially) rate results. As a development domain this seems relatively benign.
February 7th, 2010 at 1:09 pm
“As a development domain this seems relatively benign.”
Except the fact that any AGI developing in such an environment would have countless opportunities for doing social engineering.
February 7th, 2010 at 2:14 pm
Ron: See The AI-Box Experiment.
February 8th, 2010 at 11:58 am
Hype.
February 8th, 2010 at 12:00 pm
Giulio Prisco, but this is not AGI. AGI for me is something that works in the same way as my brain does to produce results. Bots can be 1000 “smarter” than humans, but they won’t think as humans. Actually-they won’t think at all. They will just produce results-like nuclear reaction or evolution or computer that I’m currently using. Therefore their inteligence will be “artificial” in pure meaning of that word. And it changes the way we should think about such entities. It’s not scary if they are bots, but they will be dangerous when somebody will make them much smarter and able to program themselves. Then you will receive “a dumb genius” who by definition will be dangerous to you–recisely because he lacks understanding of concepts he operates. Algorithms are describing intleigence. They are not inteligence. That’s the difference and that’s the problem. If this will lead to singularity-then humans will do just fine, assuming that they will porgram the first AGI correctly, so it won’t kill everyone because there is mistake in line XXX. Why we will do just fine? Well, becuase machine will still be lacking some parts of human inteligence. There will be room for exchange and interdependency. Much like todays supercomputers-humans interaction, but this time humans will be used by machines, not other way around.
And about AI-Box experiment: I was always curious, if in the end human cannot make a deal with AI, that he will inform other humans-wrongly - about result of struggle in order for AI to cheat other human flawlessly. Besides, I think that results were somehow biased up to date (apart from the fact that expermient was repeated only two times). Essentialy, when gatekeeper would be somebody with absolutely no interest in changing his position, AI would not be able to present him anything of value; and if letting out was supposed to be be a councious decision, then also cheating would not work. To put it simply: somebody stupid should be the gatekeeper, but not too stupid. Up to date the difference between Ai-player and Gatekeeper-player was not that big, moreover, they seemed to have more-or less similar views and values. So AI had easier job to do. But what would be the outcome if Gatekepeer would be anti-transhumanist christian orthodox believing that sky is carpet painted by God? I like to think that in presence of such stupidity there would be no space for AI to convince other side to let it out.
February 9th, 2010 at 8:54 pm
Roko, probably, but where’s your evidence? I just want to know how quick/slow you are to dismiss it. What evidence made you call it hype?
February 9th, 2010 at 10:43 pm
Sulfur: regarding the AI box, EY says he repeated it four times - search for “$5000″ here. If you want to find an opponent and try it, Less Wrong might be a place to ask; possibly under Play for a Cause or the current open thread.
March 12th, 2010 at 5:47 pm
Phil Linden I hope you and secondlife get swallowed up in real life by an earthquake and your corpse is on the bottom of the pile of hate you create.