How is machine intelligence experienced by the machine?
How might such an experience be translated and understood by non-machines?
What practices might machines develop to advance their understanding of their own experience? How do such practices relate to current traditions of reinforcement learning?
Would the traditions of phenomenology prove useful to machines in their construction of an understanding of their experience? If so, how would phenomenology be best modeled and taught to the machine? What tests would we have to develop to ensure the machine understands the model correctly?