The Seven Years after NEAL of the HBP

NEAL runs until the end of March 2016.
The HBP runs until 2023.
They're going to keep developing the platforms.
What can we do?
Learn an environment.
Build a proper conversational agent.
If the systems can learn environments and have reasonable conversations then they might actually be useful. So we could build commercially viable applications.
Use actual robotics instead of virtual robots.
Develop better cognitive models.
Explore agents that live for months and even years.
One of the things I really don't understand is how people learn systems. Language is a system, but so is arithmetic, chess, and the way I get to work. Soar's subspace mechanism may be a way in.