![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Agents will save the world
May. 23rd, 2007 01:54 pmMy recent conference trip was to AAMAS 2007 (which stands for Autonomous Agents and Multi-Agent Systems). This was my first agents conference so I was quite interested to find out what the rest of the field is like (outside my particular corner which involves verifying that NASA robots can pick up rubbish on Mars*). For the uninitiated, agents involve thinking about your computer program as a collection of mini-programs that can be co-ordinated to do what you want - particularly useful of course when (as with teams of robots) you really are dealing with a selection of mini-programs.
What quite surprised me was that a number of senior figures in the field believe that agents are going to save the world - although they wouldn't put it quite like that. The most concrete example came in the second keynote talk by Jeffrey Kephert of IBM. His team had developed a system where agents were used to regulate the allocation of resources (i.e., CPU cycles) in data centres. I've forgotten the exact figures but I think that if energy saving were included as a desirable quality in the negotiations then they could actually improve power efficiency in such centres by 15%. He added that if all the data centres in America could reduce power consumption by 30% it would equate to a 1.7% reduction in world-wide carbon emissions. This work has not yet be turned into a product, but coming out of IBM it isn't impossible that this could happen.
Agents are not viewed, however, as simply a technological solution to the greenhouse effect. They are also used in "social simulation" in order, in some cases, to develop policy. It wasn't clear if anyone was actually working on this in relation to energy consumption, but it was clear that some people, at least, thought that agent modelling would have a part to play in the development of policies to reduce energy consumption.
So there you go. In my previous corner of Computer Science our biggest ambition was to stop airplanes falling out of the sky (and, this being the way of things, to make sure missiles hit the things they were pointed at - as opposed to, say, civilians). The agents people have clearly set their sights much higher than that.
*at least I hope the NASA robots will feature in a concrete fashion at some point since I need to compete with the inherent cool factor of B.'s dinosaurs.
What quite surprised me was that a number of senior figures in the field believe that agents are going to save the world - although they wouldn't put it quite like that. The most concrete example came in the second keynote talk by Jeffrey Kephert of IBM. His team had developed a system where agents were used to regulate the allocation of resources (i.e., CPU cycles) in data centres. I've forgotten the exact figures but I think that if energy saving were included as a desirable quality in the negotiations then they could actually improve power efficiency in such centres by 15%. He added that if all the data centres in America could reduce power consumption by 30% it would equate to a 1.7% reduction in world-wide carbon emissions. This work has not yet be turned into a product, but coming out of IBM it isn't impossible that this could happen.
Agents are not viewed, however, as simply a technological solution to the greenhouse effect. They are also used in "social simulation" in order, in some cases, to develop policy. It wasn't clear if anyone was actually working on this in relation to energy consumption, but it was clear that some people, at least, thought that agent modelling would have a part to play in the development of policies to reduce energy consumption.
So there you go. In my previous corner of Computer Science our biggest ambition was to stop airplanes falling out of the sky (and, this being the way of things, to make sure missiles hit the things they were pointed at - as opposed to, say, civilians). The agents people have clearly set their sights much higher than that.
*at least I hope the NASA robots will feature in a concrete fashion at some point since I need to compete with the inherent cool factor of B.'s dinosaurs.