AI Summit Panel

This last few weeks we were over at GDC as part of the AI summit. We had two full days of AI discussion which was pretty wide ranging in content. Phil did two presentations, one on social behavior and another on some “tales from the trenches” under the name “Little big AI”. The big takeaway from the GDC as a whole, is that games are diverse in nature. On the high end, we saw presentations about using driven ragdolls (powered by Havok physics and animation products) in Just Cause 2. On the other end of the spectrum, we saw many indie games on the IGF stand and plenty of social games in attendance (particularly interesting was Daniel James’s talk about Three Rings facebook game experiments).

So we’re left with a feeling of wanting to work on high end 3D animation and low end flash-based facebook social simulation!

Having a quick poke through the Facebook API and how we can tie flash into it (via adobe flex) seems like it would be fun to experiment with. So its likely we will do some work on that. but will also continue work on the behavior tree code library and work on pushing box mental in a new direction. Box Mental is our 3D game/experiment and following some feedback from Sony, we decided that the concept was too hard to explain and understand as it was. It feels like we need to join the dots somewhere and actually work on something that does both high-end 3D and low end flash in a way that works with the same core gameplay and backend infrastructure. We have a background in network and distributed systems code, so it feels natural that we will bring some of that into play. We are considering how we might use either google’s appengine and/or amazon EC2 to power the backend. The key of course is to find a simple mechanic that works across a range of platforms, has a core online element but that works scalably and asychronously and that still has interesting complexities that can be controlled by AI direction.

More on all this once we have some ideas prototyped out.