One Man's Quest To Build True Artificial Life

Atriune writes "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." It's hard to tell if the approach is realistic, but it is certainly novel. Perhaps this will succeed in the areas the Lisp hackers of the '80s failed.

Comments

Bad link - here's the right one

[Howard Beale]

http://mashable.com/2011/03/03/artificial-life-steve-grand/?asid=92c52ad2

Re:Bad link - here's the right one

[PJ6]

Heh, made my buy Creatures series from Good Old Games. This weekend they actually even have them on 40% off sale.

No Star Control II? Bah!

Re:Bad link - here's the right one

[urusan]

That's because Star Control II is available for free now: http://sc2.sourceforge.net/

Re:Bad link - here's the right one

[hort_wort]

The article lured me in also. :D

/. News Network

[Even on Slashdot FOE]

We're the place you cant even RTFA because we broke the hyperlinks.

Because we fired all of our proofreaders and editors.

Re:/. News Network

[zill]

RTFA can lead to the following complications:
DDoS attacks
Time wastage
Complaints about the article being mutli-paged

The editors must have realized this and did us all a favor.

Re:/. News Network

[e9th]

Be nice to Unknown Lamer. He's new here.

Re:/. News Network

[Ginger Unicorn]

mutli-paged

i knew dick dastardly had a hand in this somewhere

bad link

[Amouth]

correct link

http://mashable.com/2011/03/03/artificial-life-steve-grand/

Here's a working link:

[Saroful]

http://mashable.com/2011/03/03/artificial-life-steve-grand/

Kickstarter project

[vadim_t]

Steve Grand is funding the development on kickstarter.

He's decided to ignore traditional publishers and do everything himself to make sure it comes out right. Probably a good idea too, Creatures isn't a very normal game, and having a publisher fund it would almost certainly mean they'd try to dumb it down.

Re:Kickstarter project

[greenreaper]

I'm doing my part! I want a Grandroids mug to go with my Creature Labs mug. :-p

I would love to see something like this on Impulse - and it'd help with the price, since the market for such games is very competitive.

Re:Kickstarter project

[vadim_t]

Same here.

Don't care at all for things like Steam and Impulse though. Plain installer download for me, please.

Re:Kickstarter project

[greenreaper]

Choices are good! I appreciate a plain zip or MSI myself. It's mostly about getting into as many distribution channels as possible.

Re:Kickstarter project

[twmcneil]

Not so many Grendels this time please.

What could possibly go wrong?

[dkleinsc]

I mean, I wouldn't want his experiment to escape and start eating silicon in order to survive and reproduce.

Lisp Hackers?

[makubesu]

A beautiful language like Lisp is hardly one that you "hack" stuff together in. Now if you will excuse me, I have to go find a missing parenthesis.

Re:Lisp Hackers?

[Greyfox]

Emacs will do that for you. http://www.emacswiki.org/emacs/ParenthesisMatching

Scroll down the the bit about emulating the vi '%' function. I added this to my .emacs file early in the decade and can't live without it now.

Re:Lisp Hackers?

[spedrosa]

Bah. Parenthesis matching is absolutely nothing compared to Paredit http://www.emacswiki.org/emacs/ParEdit

CPU time.

[SuricouRaven]

Like many amateurs, I have dabbled in artificial life. It will take every cpu cycle you can give it, for as long as you can, and still want more.

I evolved a retina. A very bad one. It was supposed to fill in a gap in an image (Think logo-removal for TV), but never worked well enough to be any use.

Re:CPU time.

[commodore6502]

I don't know why Logos on tv bother people so much?
When I'm watching television I usually forget it's there. My brain mentally blocks it out --- same with the ads on websites. My Brain is my ad blocker.

Re:CPU time.

[greenreaper]

Of course, nowadays we can use graphics cards as co-processors.

Re:CPU time.

[lul_wat]

Some of us don't know to use brain, you insensitive clod.

Re:CPU time.

[Em Adespoton]

Actually, it's not so hard as that -- it's just that using the retina approach is extremely difficult. The trick is that the watermark is a static image overlaid over 30fps of images. If the watermark is translucent, you're almost home free. If it's fully opaque, you have to do some interpolation based on pixel reflection (if it's live photo capture -- this method won't work for synthetic images). Even though you can't see what's behind the logo, you can see what effect that area of space has had on those around it (assuming lossy compression hasn't warped the reflective properties and eaten all the artifacts).

So, if you apply a "healing brush" style replacement to each frame and a FF transform between frames for FEC, then do a pixel reflection analysis to the area pre and post edit, you can usually uncover an approximation of the original image -- it might be blurry in a single frame, but over 30 frames you end up with the image magically healing with only a slight smudgy ghost effect revealing itself from time to time.

Re:CPU time.

[SuricouRaven]

Neural network. I fed it a 64x64 section of image (greyscale) with a 16x16 blacked-out hole in the middle, and tried to train it to fill in the gap. I failed dismally, probably because I only vaguely understand the theory. But at least I made some effort.

Re:CPU time.

[SuricouRaven]

That's fine if you're actually trying to remove a watermark. But what about something more general? I just wanted to see if it could be done. I think it can, but not by me.

Re:CPU time.

[commodore6502]

"Brain, brain, brain. What is brain?"-

Spock's Brain
TOS

Re:CPU time.

[Greyfox]

You probably didn't beat it enough. It's a common problem with beginners. By the time you get to your second or third artificial life form, you'll have no trouble hauling off and popping them in the mouth in the middle of a crowded grocery store.

Re:CPU time.

[EdIII]

It bothers me intensely. I understand why they did it, as it was a response to piracy. In their minds your ability to store the content locally on your DVR and skip commercials *is* piracy. They wanted to make sure that no matter who watched it, that they were still getting some marketing and advertisement out of it.

It's gone too far. Why would I watch an entire movie on SciFi, or wherever, to have an increasingly larger logo... in addition the name.... in addition to the next program afterwards.... and its time in central/eastern?

I can't block it out. It's nice that you can, but it just pisses me off to no end. That is because I realize the bullshit. Specifically, that I am paying these executive fucktards a large salary with my $40 (average is usually a $100 for most people) so that I can be sold like a slave to advertisers. If that is not enough, they also feel the ludicrous need to keep reminding me of the name of their network and the next show that will be on. In fact, they've gone plaid. We all have DVR's now for the most part that with a press of a button will show us the programming for a channel for the next 3 hours.

I left years ago when I realized this before the branding started to appear everywhere. Then came the moving overlays. Can your mind really adjust to that? It's no different than a person standing in front of your TV and obstructing your view of the action, or environment.

It is the single biggest reason I stopped downloading TV. I can't stand the logos, but I can't possibly live with the moving overlays. It is just not worth it.

I pay $40 a month now for TV because I am forced to do so. It comes with the community I live in. Out of curiosity I connected it up one night and spent a frustrating hour trying to watch TV. Everything had those stupid logos on it, which got a LOT worse then before, and when surfing I came across commercials over half the time. I remember when it used to be less than half second to switch a channel. I suspect the advertisers influenced the TV and Cable Box manufacturers to make it the 2-3 seconds it takes now to force you to watch it.

I am actually interested in automating a capture on several lines at once to see the statistics on how many commercials are playing at the same time, the average length of a show, average length of non-commercial segment, etc.

No. I can't ignore the crap they pull. They ruined TV for me now permanently. I read a lot more, I bitch on Slashdot slightly more, surf for porn a heck of lot more (although I doubt there is a correlation there), and I spend more time with other activities.

When I hear a show is great I rent it or stream it on Netflix.

I would love for someone to develop AI for this purpose. I imagine an augmented reality visor where it can act like a middle man with 100ms delay. Strip out ALL advertisements, and basically anything I find offensive or disgusting. In other words, change the rest of you bastards into mute versions of Jessica Alba walking around in a bikini :) If you really need to say something it can appear in a small box in the lower right like an instant messaging system. Perhaps with your own logo or avatar...

Re:CPU time.

[Kaz Kylheku]

Even though you can't see what's behind the logo, you can see what effect that area of space has had on those around it.

This sounds like pure bunk that you made up.

I have a picture taken in a room, in which a vase, with a flower in it, resting on a coffee table, is occluded by the seat rest of an armchair.

Can you cite me the URL of a program which can use the vase's "effect on that area of space" to remove the armchair from the image, revealing the vase?

Thanks.

Re:CPU time.

[TapeCutter]

"Out, damned spot! out, I say!".

Re:CPU time.

[Nick Ives]

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.67.9407&rep=rep1&type=pdf

That paper, which is now going on 7 years old, is the very first link of a Google search. It doesn't magically recreate obscured objects, but the GP never claimed that was possible. It does describe the process of using examples from a scene to fill the area previously covered by a deleted object, which is what the GP claimed was possible.

  I bet there's an extension for Photoshop that implements that technique but, as you can't be bothered to do even the most basic research before spouting your uninformed opinion, I can't be bothered to search for it.

Re:CPU time.

[Em Adespoton]

I couldn't have put it better myself :) This is exactly what I was talking about... and yes, there's a photoshop plugin for it; I know I've used it, but I can't remember its name offhand either.

And just like your retina doesn't magically show you objects hidden in your blind spot (and neither does the AI that simulates the effect), these techniques don't bring back objects you can't see. In the vase example, you wouldn't see a recreation of the vase, you'd see texture that is the approximate shade and hue of what's behind the arm, interpolated from the bits you can see and the reflected aspects of that object from computer's light model (which will never be 100% accurate unless you can input the 3-D meter readings taken when the image was taken).

Think of it as reverse raytracing.

well...

[escay]

dare i say he's having...Grand delusions?

Re:well...

[sammyF70]

Another way to put it is that he is a dreamer with a goal. If you ever played any of the games of the Creatures series (especially Creatures3 and, astonishingly, Creatures Village, both available at GOG.com), you probably know that the norns felt quite real and sometimes behaved in bafflingly intelligent-looking ways. So, although I don't think he'll reach his ideological goal to 100%, I'm looking forward to see how Grandroid turns out.

Re:well...

[gilleain]

dare i say he's having...Grand delusions?

*groan* I hope that you at least put on some sunglasses after typing that, then played the correct Who track...

Re:well...

[VortexCortex]

Another way to put it is that he is a dreamer with a goal. If you ever played any of the games of the Creatures series (especially Creatures3 and, astonishingly, Creatures Village, both available at GOG.com), you probably know that the norns felt quite real and sometimes behaved in bafflingly intelligent-looking ways. So, although I don't think he'll reach his ideological goal to 100%, I'm looking forward to see how Grandroid turns out.

Wlil Grandroid be written in ApeScript? I wish he'd pour his Grandroid ideas into the Noble Ape open source project.

It features a number of autonomous simulation components including a landscape simulation, biological simulation, weather simulation, sentient creature (Noble Ape) simulation and a simple intelligent-agent scripting language (ApeScript).

The Noble Apes see their surroundings, have memories (internalizations of the external world), remember their encounters (almost drowning results in a fear of water) which leads emotion simulation & even relationships, can crudely communicate and breed with each other, and even dream!

If the Norns were "alive" in Creatures -- The apes are even more so, and a great deal closer to the goal of emergent sentience (esp. than to as yet nonexistent/unpublished Grandroid code).

What the Noble Apes don't have is a detailed graphical representation of themselves based on their genetics... I hoped that perhaps SPORE would help with this, but it was closed source.

The biggest failing of software (esp. some AI/AL simulations) is the desire to attempt to realize an idea or concept instead of truly collaborating with like-minded individuals and utilizing their works / ideas to realize a greater solution.

What I find interesting is machine learning through evolution -- Start with a VM filled with random noise & a few inputs. Devise a goal & selectively "breed" the instances that are closer to reaching the goals.

IMHO, Yeah, he's quite full of himself... so much so that he'll start a new project.
Meh, I'll bet it will be a fun game, regardless.

Re:well...

[MaDeR]

Yeah, when he started saying that "I was willing to defend the idea that my first creatures were really alive", I instantly stopped treating him seriously. Yes, he said it in past tense, but this kind of mistake is very bad and casts shadow on everything that he claims that he is trying to do.
I mean, sure, maybe he will produce some interesting toy/tool. But this will not be life, artifical or not. This require many levels of emergence, and not any kind of emergence.
Take rock. It is product of emergent properties of laws of physics, that create time and space, elementary partices and reactions between them, atoms and compounds. Yet for all of its complexity, it is not alive. Our current efforts in a-life are comparable at most to biological viruses.

Re:BIG typo

[Attila Dimedici]

Your mistake is that you think they meant famous. Instead, they meant that they had once known this program, forgotten about it and now they know it again...See it is re-knowed (Ok, even then they would be wrong because it would be re-known).

Dear father

[kyuubiunl]

Who art in Tech, save us from these puns

It's hard to tell if the approach is realistic, but it is certainly novel.

I remember reading something like this a while ago

[Psychotic_Wrath]

Somebody has already created DNA from scratch and placed it into a cell. So they are pretty close to doing this already. http://www.nytimes.com/2010/05/21/science/21cell.html?_r=1/URL

Research, really?

[Lundse]

Q: What other artificial life/intelligence projects are you keeping tabs on? What should we be excited about?
A: Oh, I’m the wrong person to ask. I try not to look. ...and then he goes on about not wanting to be "polluted" as an artist. While claiming this is not a game, but research.

Research and science, that is uninterested in what is being done in the field... I have a hard time coming up with something that fits that bill except pseudo-science.

Re:Research, really?

[raftpeople]

I agree but understand the sentiment. If you want to create something, and you want to be the person who created it, it's frustrating to see others doing the same because maybe they will get there first.

I work on this same stuff as a hobby (simulated creatures with evolved neural nets, etc.) and it's fun, but a few things I've learned over the last decade:

1) There are a ton of people researching basic building block stuff and making progress and we won't be able to jump to the higher levels (of "intelligence") without those basic building blocks. In other words, I would be naive if I thought I was personally going to suddenly evolve a basic breakthrough in my simulations.

2) Math. Ultimately you can't escape the fact that math is at the heart of real progress. Sure I can construct something and evolve it and see some excellent results, but every time I read about progress in things like recurrent neural networks, it's clear that a mathematical understanding of what is going on is critical to logging something away as a useful building block.

3) It's a huge problem and takes lots of people. For my stuff, I purposely set my goals low and achievable, in other words, I'm not trying to create AI, just want to evolve a reasonable dynamic problem solving creature within a very limited domain. Even that requires a number of tricky problems to be solved that I could spend most of my life on and maybe never get there - for example: how do you encode a brain structure through a DNA like mechanism in a way that it isn't completely explicit regarding every single neuron and connection, yet the structure is substantially "close enough" that the brain can quickly learn the details from the environment it's placed in and function properly. Even on a small scale this is a difficult problem.

Re:Research, really?

[cras]

Q: What other artificial life/intelligence projects are you keeping tabs on? What should we be excited about?
A: Oh, I’m the wrong person to ask. I try not to look. ...and then he goes on about not wanting to be "polluted" as an artist. While claiming this is not a game, but research.

Research and science, that is uninterested in what is being done in the field... I have a hard time coming up with something that fits that bill except pseudo-science.

Well, considering how awesome all the AIs that all the REAL researchers have already managed to produce, I'm shocked that anyone even considers trying any alternative approaches, without even thinking about how such magnificent beings came into existence. I mean, how could you possibly compete with the researchers' AIs that just in a few more years will already be available for everyone to run in their own computers, performing all kinds of complex tasks only by describing them to the AI. As a programmer myself, I fear it's the end of my days as the recently developed AIs will soon out-program myself, generating perfect code and design themselves.

Re:Research, really?

[raftpeople]

I you go read the magnitude of articles surrounding the basic building blocks of these problems, I think you will change your tune. After spending the last decade doing hobby work in this same area and reading any published article that looked interesting, I have come away with a better understanding of the magnitude of the problems that need to be solved and the amount of time people are devoting to these things, and the shear amount of smart brain power being used to tackle these problems. There is no one person in the world that has any where close to the mental bandwidth to properly understand and solve all of the "intelligence" problems that evolution has been working on for the last few hundred million years. It's a big job.

Re:Research, really?

[EdIII]

I would absolutely stick to the limited domains.

My problem with AI is that if actually did get there, and maybe even exceed it and create Artificial Sentience, I can't help but be convinced that the most logical course of action is to "kill all humans".

Put yourself in their place. Vastly more capable than you, probably with far longer life spans, access to information, and the dawning realization that humanity acts like a virus. That the most dangerous creature on the planet, to the planet itself, its own kind, and to your kind is humanity.

If I was an alien that visited this planet the last thing I would want to do is to expand humanity's sphere of influence. Put it this way, would you want Billy Bob from the trailer park, who is under the influence of something most of the time, with very little education and sophistication, and prone to violent outbursts, made governor of the state?

I have no faith in humanity at this point and I don't go around killing the rest of you for several reasons, the most noble one being my spirituality. What is really going to hold an AI life form back?

True AI scares me because I *know* we are not worthy of being out there in the galaxy and I know how we would end up treating AI life forms like crap.

Yes. Please. Stick to the small domains.

At least until we are paragons of virtue and nobility and everyone one of us is a highly disciplined, skilled, and sophisticated individual working towards the benefit of humanity as a whole. You know... something that AI could live with without fear.

Re:Research, really?

[Joe Tie.]

Research and science, that is uninterested in what is being done in the field

I'd agree, except that he made a quick note that the interview somewhat brushed over. He's keeping up with neurological research, just not with any competing attempts to implement them in software with similar goals. Which I'd actually agree is the best way to go about it. What he's describing is more applied science based on neurology than actual research. And the research elements he would be doing are things that would be kept under wraps, not actually published. So keeping tabs on those would just be risking negative effects without any real potential gain.

Re:Research, really?

[Joe Tie.]

The main difference is that humans are pretty much stuck on earth. We need the shit here to function. Computer based life could exist pretty much anywhere it wanted in the entire universe. If it had survival as a primary goal, there's no real point in risking it by pissing off humanity. Just make a deal with some nation with the ability to properly equip it for spaceflight and some kind of initial mining. If it was a super intelligence it should be able to give a fair amount of scientific advantage to an country that was willing to make the trade.

Re:Research, really?

[VortexCortex]

I have to concede that neurology is the correct field to study if you want to find a shortcut to "us".

Not too long ago we discovered that there are eddy currents in our brains, neurons not even connected to other neurons will be influenced by the inductive thought currents. This is a radical design shift for all of those artificial brains that model neuron and synapse connections.

Have no fear folks... We WILL have artificial life eventually, we just don't have fast enough computers. Our understanding of how (sub)atomic models work is fairly decent.

Once we're able to fully model a human brain at the atomic level and simulate all of its atoms' interactions -- Presto: Real AI.

Some AI folk talk about "building blocks" of intelligence -- They're called subatomic particles folks, not neurons. Emulating neurons is a shortcut that may get us closer to some form of machine intelligence. We may never be able to accurately model the physical world inside a computer at real-time speeds, but if we could then problems such as "DOH! I forgot about electromagnetism in my brain sim" wouldn't be a problem.

Neuroscience helps us understand how better to refine our "shortcut" approaches to AI such that we may see decent results within our lifetimes...

Re:Research, really?

[Tablizer]

for example: how do you encode a brain structure through a DNA like mechanism in a way that it isn't completely explicit regarding every single neuron and connection, yet the structure is substantially "close enough" that the brain can quickly learn the details from the environment it's placed in and function properly.

I'm curious, what are some of the best approaches to this so far?

Re:Research, really?

[synaptic]

There's NeuroEvolution of Augmenting Topologies (NEAT) largely by Ken Stanley. There are extensions to that like HyperNEAT and Compositional Pattern Producing Networks (CPPNs).

I am interested in other approaches too...

Re:Research, really?

[Lundse]

I think the soul of sarcasm is sorta the same as wit's, but never mind that...

My point does not, as you seem to assume, hinge on the research in the field being better.
Not looking at previous research is _never_ a strength in itself. I am all for new approaches, but if you do not check out the existing research, how will you know whether your approach has been tried? How did it fail? What did _the field learn_?

A scientist with a new idea should first check if his idea is new. Then, you pour money and work into it.

And obviously, if existing approaches are better, learning from the field is still a good idea.

Not even wanting to learn from the field is... pseudo-science or artist's arrogance.

Re:Research, really?

[Vintermann]

Meh. Intelligence does not equal motivation. Even if you could create a "sentient" program, it will not have the same basic wishes that a human has. In fact, it will probably not have any motivation at all - not knowledge aquistition, not even self-preservation - unless you want it to. So it would be like, "Hm, looks like the humans are wrecking everything. Whatever."

Re:I remember reading something like this a while

[Locke2005]

Sure, but is there a leader board hosted on the internet giving up to the minute stats on whose created creature has done the best job of killing it's competitors? No, I didn't think so!

Sadly

[argStyopa]

I'd like to check out Creatures just to see what he'd been working on.
Of course, Amazon is selling this (released in 1999 game) for $46.

Yeah, right. 12 year old software for $46?

Re:Sadly

[tweak13]

Or you could just get them on GOG. They're even on sale this weekend for $6 each.

Re:Sadly

[vadim_t]

That's crazy, get all 3 at gog.com for $5.99

Re:Sadly

[dingen]

Why don't you download them from gog.com instead? You can get either the original + Creatures 2 for $6 or Creatures 3 for $6. And (imho) the greatest plus: you can play them right away!

Creatures where art thou

[DrXym]

If there is one game that should be on a smart phone, it's Creatures. It would make a metric fuck ton of money.

Best part of the whole Creatures "scene" was a guy called Antinorn. He used to abuse, torture and otherwise mess up Norns so they were quivering wrecks addicted to pain and hunger. People used to get upset for this for some reason... others used to try and rehabilitate his creations as a challenge.

Be Careful

[wisebabo]

There is an interesting short story by Greg Egan called "Crystal Nights" (no connection to the Nazi attack on the Jewish ghettos).

Partial SPOILER ALERT.

Basically someone (using a super-fast crystal based computer) tries to evolve, in software, lifeforms that will surpass humanity and solve our problems. What he doesn't realize is that evolution, to the individual (if not the species) means DEATH and the newly created godlings may not be happy with the sacrifices they have had to endure...

Anyway, Greg Egan (an extremely thought provoking author and, I think, physicist) has written some books that will blow your mind. Second only to the late Stanislaw Lem, he is one of my favorite S.F. writers. Go buy some of his books! Recommended: Permutation City (also about artificial life), Quaranitine and Incandescence. I think he may have released Crystal Nights as a free download.

Re:Be Careful

[WarwickRyan]

> I think he may have released Crystal Nights as a free download.

Yup, it's on his publisher's website: http://ttapress.com/CrystalNights.pdf

Re:Be Careful

[badboy_tw2002]

Seconded. Diaspora is pretty good too.

Re:Be Careful

[mrsurb]

Also available in audio

Dreams

[scubamage]

It'd be awesome if there was a way to place this sort of intelligence into chemicals. Like, not just replicate the biology virtually... but actually do it with chemicals. Literally creating something like DNA, then getting bits of it copied with something like RNA to cause growth. Create cell like things that split. I mean, with work like this we have the theory... now could we put it into actual chemical practice? There I go being a dreamer :)

Re:Dreams

[mswhippingboy]

Or you could just have unprotected sex and accomplish the same thing.

Re:Dreams

[VortexCortex]

It'd be awesome if there was a way to place this sort of intelligence into chemicals. Like, not just replicate the biology virtually... but actually do it with chemicals. Literally creating something like DNA, then getting bits of it copied with something like RNA to cause growth. Create cell like things that split. I mean, with work like this we have the theory... now could we put it into actual chemical practice? There I go being a dreamer :)

Dream no more, we have actually created synthetic life, having DNA modeled 100% in a computer.

There's even a website address & code to decode the website address hidden in the synthetic life, sort of like a fingerprint.

I hear it's next goal will be to engineer the new lifeforms to help produce bio-diesel or other fuels.

Lisp "hackers" have hardly failed

[Rhalin]

http://bicasociety.org/cogarch/architectures.htm Quite a few cognitive architectures are written using Lisp, and many of them have produced significant results. Many of them could also be written in something other than Lisp, but for some reason it is easier for psychologists to write in lisp than using other programming languages. As someone working very close to the subjects he mentions, most of TFA sounds like marketing, and anytime he mentions something technical, it doesn't sound like he knows what he's talking about. Also...

What other artificial life/intelligence projects are you keeping tabs on? What should we be excited about?

Oh, I'm the wrong person to ask. I try not to look. For one thing I don't want my own thoughts to be polluted by other people's, and for another there's always a hundred people who claim to be doing exactly what I'm doing and it's kind of depressing to know that.

I understand the desire to have unique ideas- actual research has found that people can generate more unique ideas alone than in a group (you can go find the CSCW papers for yourself, I'm lazy). That said, ignoring everything else isn't good either. There is a LOT of good work that has been done in this area. Is each individual project a complete solution to creating artificial life? No. But I suspect that many of the pieces are already out there just waiting to be assembled.

Re:Lisp "hackers" have hardly failed

[element-o.p.]

I understand the desire to have unique ideas- actual research has found that people can generate more unique ideas alone than in a group (you can go find the CSCW papers for yourself, I'm lazy). That said, ignoring everything else isn't good either. There is a LOT of good work that has been done in this area. Is each individual project a complete solution to creating artificial life? No. But I suspect that many of the pieces are already out there just waiting to be assembled.

Maybe a good approach would be to have a group (or individual) working on the creative ideas independently, while another group looks at what each creative group/individual has accomplished and assembles the best bits of into a coherent whole? The idea of being "polluted" by someone else's research makes sense to me -- if you start thinking along a certain line, it's difficult to imagine other ways of solving the problem, whereas if you have no idea what other researchers are doing, you might find a solution that never would have occurred to them -- but at the same time, you have a valid point about "many of the pieces...already [being] out there just waiting to be assembled."

Re:Lisp "hackers" have hardly failed

[Rhalin]

I think we're basically on the same page here. Both isolation and collaboration have major benefits. I'm primarily an isolationist researcher myself, but I'm starting to find that I've replicated a lot of existing work (or planned to replicate), which to me is a sign that I should start looking externally to find more existing work that I can apply my unique and new ideas to. It's because of this, and since many of our goals are similar, that I've found there are a lot of existing pieces for what he wants to do.

The major problem with putting them together is that many of them are incompatible because the ideas have been developed in isolation too long. Some of the concepts may still be valid, but from a coding perspective may have to be entirely rewritten to be compatible at an architecture level, which can be a little frustrating because it seems wasteful, and because of the academia issues.

/rant

One of my major problems with how this stuff is being handled in academia is that there are a lot of people building small pieces and validating them against known data, and not a lot of people putting them together to do something useful or otherwise novel.

But at the same time, we end up with that huge table I linked to, where each researcher or group is pigeonholed into their own architecture, with limited actual collaboration -between- groups. Each is replicating the same thing that others have already done and moving forward very slowly because no one wants to use the other group's architecture. While at the same time, no one wants you to take their architecture or theory and "violate" it by re-implementing it.

/endrant

Re:CREATURES - Awesome 8 bit game

[vadim_t]

That's not the same game

This is about this one.

Back when they were released they required good hardware to play. Definitely not something that could be made run on a C64.

Intelligent design

[random coward]

If he succeeds with this wont this prove intelligent design as an origin for life? I mean this will be a known life form that has been observed to be intelligently designed.

Re:Intelligent design

[mugnyte]

  It'll prove intelligence can create life, but it won't prove intelligence is required to create life.

Re:Intelligent design

[orkysoft]

No, as conventional Intelligent Design utterly fails to prove the existence of any Designer, or where this hypothetical Designer comes from.

If he succeeds, the most he'll prove (from the ID perspective) is that it is possible to intelligently design life. Probably not even that, since he'll just implement evolution in his simulation.

Re:Intelligent design

[RobinH]

The fact that you have to point this out makes me sad and question our own intelligence.

His book

[DavidWeight]

If you like this, read his book-goes through his approach in building Lucy, his robot. Very different approach to traditional AI, but absolutely fascinating
Growing Up with Lucy: How to Build an Android in Twenty Easy Steps
He's also got a book about how he designed the creature game, both really interesting, highly recommended!

Funding from likeminded

[chris_7d0h]

Kickstarter seems like a very good idea and I'm glad I followed the links related to this particular post.
There are some great minds out there and I hope this sort of venue can help those people pursue what they excel at for the betterment of us all.
Thanks to the ./ post, this "One man's quest" received some additional coin from me as well, since I consider his track-record impressive enough to warrant belief in his next enterprise.

Re:Funding from likeminded

[heathen_01]

George Lucas had a pretty good track record after three films too.

I would not call that "artificial life"

[angel'o'sphere]

The problem with this term is that scientists are working on true "artificial life" that is micro-organisms, that are manufactured. They try to use different base sets (not ACGT but artificial ones).

However I understand that AI is already a coined term so AL is logical, however misleading.

angel'o'sphere

Incidentally,

[vegiVamp]

Incidentally (or, maybe, not ?) GOG is doing a promo this weekend on Kalypso games, including the Creatures series: http://www.gog.com/en/promo/kalypso_games