1.) Please tell us a little about yourself.
I'm currently in the last stages of my PhD, here at the School of Cognitive and Computing Sciences, University of Sussex, Brighton UK, under the supervision of Phil Husbands. I'm about to start a post-doc shortly.
My work is focused on the evolution of social behaviors and, in particular, social coordination and communication. I have studied both evolutionary/ecological and behavioral/cognitive models of social coordination by analytical and computational means.
Among other things, my current interests include theoretical aspects of social interactions from the point of view of autonomous systems, evolutionary dynamics and self-organizing processes, evolution of language as a biological
phenomenon and the relationships between social behaviors and cognition.
I'm also interested in methodological issues concerning the use of (ALife) computer simulations in scientific practice.
2.) Just where does ALife begin and Artificial Intelligence end?
Alife and AI are two disciplines of scientific research. Unfortunately, in contrast with other disciplines such as physics or chemistry, the aims of AI and ALife are not se clearly defined. Many researchers approach these disciplines because they have what we may call "engineering" interests. And so their aim is to build intelligent computer programs, or robots, or artificial life-like processes. Other people see these disciplines as research tools, and their interest could be classified more as "scientific". Thus, these people would use Alife techniques to address questions in theoretical biology, or use AI to test the consistency of theories of human cognition.
The differences between the disciplines will then depend on what end of the spectrum you are situated. It could roughly be said that ALife
is more interested in biologicial phenomena in general, while AI is not. However, for Alife researchers human cognition IS a biological phenomenon, therefore there would seem to be an overlap between disciplines. However, there are many differences. While AI attempts to model higher level cognitive processes (such as the ability to play chess), Alife addresses cognition in simpler, but complete, organisms (for instance, obstacle avoidance). The underlying assumption is that this methodology is much more faithful to biological constraints, while in AI, any computationally realizable process may be postulated without actually caring about biological plausibility.
3.) Will Artificial Intelligence comes naturally if Artificial Life is successfully created?
Again, this is a question where the distinction between engineering and science may be useful. In the scientific front, (where these techniques are use as tools of research) this question makes no sense, since no one is trying to "create" anything. In the engineering from, well, it all depends on your goals. As far as I know, noone is actually trying to create artificial organisms as such, but many researchers are interesting in endowing robots and artificial agents with life-like characteristics, such as autonomy, etc. If successful, this type of research shows indeed a lower level sort of artificial intelligence. It does not follow that your robot will learn to play chess after enough training or evolution. There are many issues that have not been addressed so far, such as the evolution of symbol use, that would
seem necessary to account for many human cognitive activities. So there is a gulf between the engineering goals of ALife and AI. But it does not mean that it is a gulf that cannot be crossed.
4.) Does Artificial Intelligence constitute Artificial Life?
Short answer: No. I think this follows from the answers above.
5.) What are the practical applications of ALife?
Well, many. From actual robotic agents endowed with sufficient autonomy to help them cope with unpredictable environments, to software agents that trade goods over the Net. Other applications include, entertainment, evolutionary search, evolutionary design, etc.
6.) How can ALife be applied to robotics?
By helping in the determination of what sort of cognitive architectures are used to control the robots. A great portion of the ALife field is called evolutionary robotics. The methodology consists on "evolving" control architectures (in the form of neural networks, for instance) that will help the robot in coping with certain complex task. The taks of controlling navigation and other parts of the robot's body in noisy environments is too complex to be designed succesfully by humans. So evolutionary techniques (for instance, genetic algorithms) are a good possible alternative ways of finding good solutions to this design problem.
