In an earlier posting I wrote
How about planting a 100% autonomous agent ‘out there’ in Second Life… let it roam the world, and blog back what’s happening?
Vlad has kindly pointed me to someone proposing something similar.
In a paper entitled “Creating Wholly Autonomous Agents in On-line Worlds” (from Proceedings of the 2005 IEEE/WIC/ACM International Conference on Intelligent Agent Technology (IAT’05)), Steven Willmott writes:
The aim of this paper is consider how it might already be possible (although potentially illegal) to exploit these on-line environments in order to construct wholly autonomous electronic entities able to act for themselves: sustaining themselves financially, choosing their own actions, interacting with humans – and surviving for extended periods without external assistance. The paper is structured as a “thought experiment” on how mechanisms necessary for self sustainability might be put in place and the potential implications of creating such automated entities.
It’s not obvious to me that this is illegal – I guess “it depends”! And Willmott is only at the thought experiment stage, so we haven’t been scooped yet! However, his paper is an important exploration of the territory, identifying some of the key issues of identity, autonomy, situatedness, mobility, legality, and setting out potential criteria for success.
My conjecture is that this arena (autonomous bots in Second Life and any of numerous other environments) will turn out to be as influential as the Turing Test as a great way to judge the performance and behaviour of software tools in an otherwise human or proxy-human context.
Technorati Tags: Second, Life, SL, bot, autonomy, autonomous, intelligent.agent, Willmott
January 25, 2007 at 1:44 pm
Are there any open source AI NPC game engines out there that could be repurposed for this. Maybe add a dash of jabberwacky, too?
tony
January 26, 2007 at 7:57 am
Worth checking out – would love to see some of your Lego robot programming tools applied to SL bot programming for kids…
January 30, 2007 at 12:03 am
[...] Therefore, as in the the paper blogged by Marc in which an autonomous agent survives on a server by earning money in virtual environments to sustain itself for ever, the interesting bit is the importance of actions over existence and reflection. Even if not clever the agent is believed to be human “most of the time”. I believe with immortal computing people will believe we are alive “most of the time” when they will get news and information. [...]
January 30, 2007 at 12:33 am
[...] Therefore, as in the the paper blogged by Marc, in which an autonomous agent survives on a server by earning money in virtual environments to sustain itself for ever, the interesting bit is the importance of actions over reflection and even, in this case, life. Even if not clever the agent is believed to be human “most of the time”; with immortal computing people will believe I am alive “most of the time” when they will get news and information, depending on the amount and the smartness of our residual activity, forgetting for a moment that no one is there anymore. [...]