A novel approach to Artificial Life

For the discussion of the sciences. Physics problems, chemistry equations, biology weirdness, it all goes here.

Moderators: gmalivuk, Moderators General, Prelates

SpaceShipRat
Posts: 55
Joined: Sat Jul 10, 2010 8:18 pm UTC

A novel approach to Artificial Life

Postby SpaceShipRat » Tue Mar 29, 2011 2:31 pm UTC

"The creator of the renowned Creatures artificial life series is at it again. Fifteen years after the initial success of the Creatures Trilogy, Steve Grand continues his quest to go beyond simulation, and create real artificial life."

http://mashable.com/2011/03/03/artifici ... qus_thread

I already posted in the games section, but I'd like to see some opinions from the science, rather than game pov.

Moose Hole
Posts: 398
Joined: Fri Jul 09, 2010 1:34 pm UTC

Re: A novel approach to Artificial Life

Postby Moose Hole » Tue Mar 29, 2011 3:18 pm UTC

Sounds cool, but how much brain cell simulation can he get out of a computer? It would be cool if he could make something like that learn behaviors, probably the easiest one to detect would be fear. It seems to me that it would be hard to know if you have anything really intelligent, it takes humans years to appear to be smarter than a dog. Maybe if he could run a lot of inputs quickly, by using computer simulations for the intelligence, that might speed things up, I guess.

User avatar
Charlie!
Posts: 2035
Joined: Sat Jan 12, 2008 8:20 pm UTC

Re: A novel approach to Artificial Life

Postby Charlie! » Tue Mar 29, 2011 3:57 pm UTC

We won't be greeting them as our overlords anytime soon. He gives the impression that his artificial life will be controlled by a neural net. Well, IBM, on a big ol' supercomputer, can simulate something about as complex as an ant brain. So the only way to get humanly interesting (i.e. high-level) responses from a simple neural net seems like hooking it up to already high-level outputs (i.e. a "smile" command, rather than a motor cortex). Now, this network's structure may be encoded in genes, allowing for some cool simulated evolution, but even if you put it on a supercomputer it'll probably just go to some local maximum and stay there, since its options are restricted by having to also be humanly interesting.
Some people tell me I laugh too much. To them I say, "ha ha ha!"

Moose Hole
Posts: 398
Joined: Fri Jul 09, 2010 1:34 pm UTC

Re: A novel approach to Artificial Life

Postby Moose Hole » Tue Mar 29, 2011 4:08 pm UTC

Charlie! wrote:We won't be greeting them as our overlords anytime soon. He gives the impression that his artificial life will be controlled by a neural net. Well, IBM, on a big ol' supercomputer, can simulate something about as complex as an ant brain. So the only way to get humanly interesting (i.e. high-level) responses from a simple neural net seems like hooking it up to already high-level outputs (i.e. a "smile" command, rather than a motor cortex). Now, this network's structure may be encoded in genes, allowing for some cool simulated evolution, but even if you put it on a supercomputer it'll probably just go to some local maximum and stay there, since its options are restricted by having to also be humanly interesting.
That's not how neural nets work. Anyway, as you say, some of our actions are dictated by the nature of how we work, such as the motor cortex allowing us to smile (though we learn why we should smile by experience). I'm guessing to make this work at all, some "instinct" level stuff needs to be there so that it can have a base to work off of. Whether this is a genetically created instinct or one that is preprogrammed in this application will be interesting to see.

Technical Ben
Posts: 2986
Joined: Tue May 27, 2008 10:42 pm UTC

Re: A novel approach to Artificial Life

Postby Technical Ben » Tue Mar 29, 2011 5:14 pm UTC

So. One guy says he has created a fully functional AI and no one questions it?

Why did I not file that letter to the news paper saying I'd invented a "teleporter"...

I'm guessing this AI goes as far as the claims that come from the likes of Will Wright and Peter Molyneux.
It's all physics and stamp collecting.
It's not a particle or a wave. It's just an exchange.

Moose Hole
Posts: 398
Joined: Fri Jul 09, 2010 1:34 pm UTC

Re: A novel approach to Artificial Life

Postby Moose Hole » Tue Mar 29, 2011 5:39 pm UTC

He hasn't created it yet, but people are paying $50 a pop to play with it in the future.

Turtlewing
Posts: 236
Joined: Tue Nov 03, 2009 5:22 pm UTC

Re: A novel approach to Artificial Life

Postby Turtlewing » Tue Mar 29, 2011 6:07 pm UTC

Technical Ben wrote:So. One guy says he has created a fully functional AI and no one questions it?

Why did I not file that letter to the news paper saying I'd invented a "teleporter"...

I'm guessing this AI goes as far as the claims that come from the likes of Will Wright and Peter Molyneux.


It helps that
1: he's made a name for himself in the field of AI
2: he hasn't actually made detailed claims about the capabilities of the creatures. His claims are basically that it will use models based on organic life rather than models based on game design principles for generating the creatures and their behavior. A sufficiently simplified model of "biochemistry" and a sufficiently constrained would would make that a much more attainable goal than the typical (layperson's) concept of an "AI".

In all honesty his claims aren't that far removed from what some people I know though Spore would be.

Technical Ben
Posts: 2986
Joined: Tue May 27, 2008 10:42 pm UTC

Re: A novel approach to Artificial Life

Postby Technical Ben » Tue Mar 29, 2011 8:43 pm UTC

And what did spore become? ;)
It's all physics and stamp collecting.
It's not a particle or a wave. It's just an exchange.

Moose Hole
Posts: 398
Joined: Fri Jul 09, 2010 1:34 pm UTC

Re: A novel approach to Artificial Life

Postby Moose Hole » Tue Mar 29, 2011 8:52 pm UTC

Technical Ben wrote:And what did spore become? ;)
A piece of crap. Did you even read the article where it says he's raising his own money so he doesn't have to deal with salespeople and focus groups screwing with his ideas? Try that.

SpaceShipRat
Posts: 55
Joined: Sat Jul 10, 2010 8:18 pm UTC

Re: A novel approach to Artificial Life

Postby SpaceShipRat » Tue Mar 29, 2011 11:52 pm UTC

Don't just write it off with the usual "AI is alway a failure" knee-jerk reaction.
I know it all sounds improbable, and I'm certainly not expecting to be able to carry out an intelligent conversation with these things. But within the scope of the thing, an ALife game -slash- collective experiment, I believe we have reason to be greatly excited.

Besides, he's not "just some guy", he's even got a Wiki article. well, stub.

Artificial Life/ Intelligence can't stay still forever, the winter always ends and gives way to spring.
Last edited by SpaceShipRat on Sat Apr 02, 2011 11:10 pm UTC, edited 1 time in total.

Soralin
Posts: 1347
Joined: Wed May 07, 2008 12:06 am UTC

Re: A novel approach to Artificial Life

Postby Soralin » Wed Mar 30, 2011 12:25 am UTC

Technical Ben wrote:So. One guy says he has created a fully functional AI and no one questions it?

Why did I not file that letter to the news paper saying I'd invented a "teleporter"...

I'm guessing this AI goes as far as the claims that come from the likes of Will Wright and Peter Molyneux.

Why would we question it? AI is a common feature of most games, in varying degrees. AI defines it as just: "a system that perceives its environment and takes actions that maximize its chances of success." (for example, basically anything in a game that has a goal(shoot you, capture the king, etc.) and reacts to what you do to try and achieve that goal, would be AI)

I suspect you're thinking that the term AI is referring to Strong AI, or general intelligence or such. That would be huge, but then that's not the claim that's being made, as AI, especially as used by people who work on it, typically just refers to the first definition.

SpaceShipRat
Posts: 55
Joined: Sat Jul 10, 2010 8:18 pm UTC

Re: A novel approach to Artificial Life

Postby SpaceShipRat » Wed Mar 30, 2011 7:30 pm UTC

According to the last update, they're also going to be learning to move about by controlling muscles in their limbs, rather than have a set of animations. I think i've seen a study of that sort done with robots, where they had to work out a gait via genetic algorythms, can anyone remember that?
Just imagine how cute the newborns will be, staggering about, tripping over their feet. Also, the guilty amusement of pushing them off balance just to mess with them XD

Soralin
Posts: 1347
Joined: Wed May 07, 2008 12:06 am UTC

Re: A novel approach to Artificial Life

Postby Soralin » Thu Mar 31, 2011 12:36 am UTC

SpaceShipRat wrote:According to the last update, Thor also going to be learning to move about by controlling muscles in Thor limbs, rather than have a set of animations. I think i've seen a study of that sort done with robots, where they had to work out a gait via genetic algorythms, can anyone remember that?
Just imagine how cute the newborns will be, staggering about, tripping over Thor feet. Also, the guilty amusement of pushing them off balance just to mess with them XD

There have been a few, either with robotics, or doing things in software, or a combination thereof. There's a simple version of something like that, although with swimming organisms, with genepool: http://www.swimbots.com/ I don't know of where to find something similar for walking creatures, but this is free. :)

SpaceShipRat
Posts: 55
Joined: Sat Jul 10, 2010 8:18 pm UTC

Re: A novel approach to Artificial Life

Postby SpaceShipRat » Sat Apr 09, 2011 12:22 pm UTC

Just thought I'd give you a heads up- time is almost over.

roflcopter
Posts: 26
Joined: Thu Nov 13, 2008 2:14 am UTC

Re: A novel approach to Artificial Life

Postby roflcopter » Mon Apr 11, 2011 6:34 am UTC

Well I skimmed most of the posts and didn't actually read any of the links, but to me artificial life and artificial intelligence are two completely different things...

That is all.

Soralin
Posts: 1347
Joined: Wed May 07, 2008 12:06 am UTC

Re: A novel approach to Artificial Life

Postby Soralin » Mon Apr 11, 2011 9:32 am UTC

Well there's that too, I mean, bacteria are life, plants are life, but I wouldn't exactly call those particularly intelligent. Although "real artificial life" seems it would better apply to something like a self-replicating machine, whereas this would probably be closer to "simulated artificial life", or "artificial artificial life" or something like that. :)


Return to “Science”

Who is online

Users browsing this forum: No registered users and 15 guests