Conclusion on CAs
- We've got a working simulator that is pretty efficient.
- We're gaining a pretty good understanding of it when it
comes to classifying one input, including hierarchies. Of course
there is still more to learn here.
- The next step is to move into understanding timing, and thus
sequences.
- A parallel track is to understand variable binding.
- We can currently start to work on having one system to
solve multiple problems.
- As we integrate variable binding, and timing, we can start
to move to a more complete architecture.
- This architecture will be cognitive and thus a good psychological
model.
- It will also solve real AI problems and so be the basis for a
real AI.
