It seems that optimization power as it’s currently defined would be a value that doesn’t change with time (unless the agent’s preferences change with time). This might be fine depending what you’re looking for, but the definition of optimization power that I’m looking for would allow an agent to gain or lose optimization power.
It seems that optimization power as it’s currently defined would be a value that doesn’t change with time (unless the agent’s preferences change with time). This might be fine depending what you’re looking for, but the definition of optimization power that I’m looking for would allow an agent to gain or lose optimization power.