This Forum is Kinda Empty...Let's Get Started

I'm always lookin for things like the SL Ecosystem, AI Pets, Advanced Chatbots, Greeters and Attachments...anything to do w/ AI in SL. If anybody has a link, or SUrl to a location where I can find a bot to chat with, or any other info on decent AI Bots in-world, why not list em here?

I just recently joined the group and was surprised to find no activity. I would have thought that sl is the perfect place to explore a lot of AI concepts. This is especially surprising with the number of technical types that hang around. Are there other groups that fill this niche that I don't know about?

I really wish I had more time to spend in SL working on my AI project, but I think I've only spent 10 mins in-world this whole year...

My project is not abandoned, I have a long list of improvements and code re-writes that I want to do. I just don't have the time right now


Surina

Also recently joined and equally surprised at the lack of activity.

Does this mean there is no AL activity/research/development going on in SL or
is everybody to busy to discuss it?

How about a thread here to collect all ongoing work in progress (even if its not much
progress ?

Well, keep in mind that LSL is VERY limited. Building anything complex is a very trying experience both because of the stated limits of LSL and the very large number of bugs and weird behaviours with both LSL and the SL environment.

So does this mean that the reason not much AL projects going on is that SL is'nt a good platform for AL?

Did many try and not succeed?

I do see the limitations (while coding my own little AL project) but I planned to offload anything to hard for LSL on a external webserver anyway. So far it was'nt necessary to go this way but we'll see.

SL is actually a very good platform for AL as it provides most of the functions for you that you'd usually have to write yourself if creating your own application from scratch such as physics, collision detection, object creation/replication, etc. This leaves you to concentrate on the functionality, ie, making your aLife creatures behave life like.

There are two problems with LSL, lack of speed and lack of storage. The speed issue can be addressed by splitting your code into several scripts that pass info to each other. Carefull selection of which functions go into different scripts will result in effective parrallel processing and greatly improve performance, eg a sensor script that passes results to another script for processing while it does another sensor sweep... Offline storage to simulate long term memory could be a usefull technique.

The speed issue should be fixed once the mono engine comes online.

Wish I had more time....

Surina

p.s. I'll try to get online sometime to fix the fish.

Hi I'm not an AI expert at all, but I did meet an interesting chatbot called ALICE in SL, at a place called Devil's Moon owned by Abramelin Wolfe (sorry I don't have a SLurl) and wrote about it in my blog ("Marion in Wonderland" http://slnewbiediary.blogspot.com/2007/01/go-ask-alice.html).

There's also info about SL chatbots and other AI stuff on www.alicebot.org

Good to hear since that were exactly the reason I chose SL for my AL projects.
And besides all other things are somewhat limited to share with others.
Using Human/Avies as evoultionary driving force is not very feasible unless there are enough paricipants.



And since there is no intrinsic include file or library writing support splitting up
plus messaging seems to be the main modularizing possibility without help of external preprocessors.

Maybe I'm wrong but moving to a plot on sparsly populated class 5 sim speeded up the scripts quite a bit.





That would be very nice ... did'nt see them so far.

I'm pleased there has been some response. I'm very interested in AI - and in AL, but don't have the time to start a programming project. I wonder if there would be enough interest in forming a group to buy a sim for the sole purpose of pursuing projects with a clean Class V server, per Prion. Whether this is or is not the ideal platform for these things from a the point of view of computational speed, the infrastructure already here for visual interection and group participation makes it very interesting.

I wonder if this is a useful idea. I have been thinking about buying an island to do something, but just haven't had the time to take on a big creative process by myself.

p.s. I'll try to get online sometime to fix the fish.[/QUOTE]

What fish?

I have an intense interest in both AI and AL, but simply don't have much time to work on it. In between RL work and the work that I do in SL to make a bit of money on the side, there isn't enough time in the day.

I did make a combat-capable semi-autonomous bot, with the goal that I would one day add something like language processing (or even just a rudimentary facsimile) as well as obstacle avoidance and pathfinding and use it as a testbed for other similar concepts (just imagine, autonomous swarming combat bots that are capable of learning!!!), but I simply don't have the time or knowledge required.

I guess it might be different if I was an expert in such things, but I am a mere bumbling amateur attempting to learn what I might about AI on my own. Someone who truly knew what they were doing could perhaps do it faster and would probably make it a higher priority than I would.

I missed all the old posts because I did'nt realize the default "Display Options -> From the" is set to "Last Month".

So there has been quite a lot of activity until about fall 2006 and then it stalled somehow.

In case other forum newbies did the same error as me, set the "Display Options -> From the" to "Beginning", I just catched up on the whole postings and it is defintely worth reading.

Really cool. Of course nature is full of complex behavior from oganisms that have limited responses. There are a series of movies showing very complicated behavior of slime molds - an example of complex behavior with zero neurons.

By adding just a little bit of complexity to the individuals you get exponentially increased complex behavior (I think). e.g. recent experiments with small robots with simple rules of interaction which evolved deceptive behavior in just 500 life cycles.

Sorry, just rambling - you all know this.

By the way, can the fish be seen anywhere now?

I have some evolving prim objects that you can write modules for if anyone is interested in contributing.

I would also like to see the fish. I just joined this group and have ambitiously set myself the project of making a hummingbird. I've made a beautiful feeder, but now I need a bird for it.

Are the fish viewable in world?

Any action in the artificial life scene lately?

I just joined second life to explore developing autonomous bots using neural network and GA techniques, which is what I do for a living with robots in first world (see www.7gt.com). I use simulation a lot in developing the robots, so second life seems very appealing and with lots of uses in second life--then some could also be exported to first life--but I'm not finding much about other people doing anything like that here. Seems like high bandwidth sensing will be an issue (inputting camera input and sound to a neural network, not simple ones like touch and text sensors), and the computation-intensive neural processing is probably a challenge, since scripts don't seem suited to this. Would a dedicated external server work? Sorry for being so naive, but it would be very helpful to find out now if this is completely unrealistic in this environment, so I appreciate any feedback from vets before I invest too much effort. I'd be happy to collaborate, also.