Conclusion
- We didn't expect to get the grant, so we implemented this
vision system in less then 40 hours of hacking.
- This implies that the neural approach has a lot of promise.
- We're hoping to get a complete working agent by next August.
- After that we'll have the ability to explore a range of
problems including learning words, learning actions,
prepositional phrase disambiguation, learning new objects,
and more.
- This has the potential to bridge the gap between symbolic
and subsymbolic AI.
- This may solve the brittleness problem of AI, and allow
domain general systems.
- We're looking for collaborators.
