Title: CABot2: a neural agent that is a step toward AI
by Chris Huyck

In the presentation, I will present CABot2, an agent in a video game
that assists the user. CABot2 takes natural language commands, makes
and executes plans, views the environment, and learns.  It does all of
this with just simulated neurons.  The long-term hope is that this
success can be extended to an implementation of a neural cognitive
architecture.  (The natural language parsing component of CABot2 is
already a good cognitive model.)  Some current problems being pursued
(conversational systems, cogntive mapping, improved vision and
improved cell assemblies) will be outlined.