Can you think of RL in an infinite/continuous state space as a machine that can generate infinite examples?

What work exists in machines that can auto design reward functions?

#ai #question