I argue that data becomes temporarily interesting by itself to someself-improving, but computationally limited, subjective observer once he learnsto predict or compress the data in a better way, thus making it subjectivelysimpler and more beautiful. Curiosity is the desire to create or discover morenon-random, non-arbitrary, regular data that is novel and surprising not in thetraditional sense of Boltzmann and Shannon but in the sense that it allows forcompression progress because its regularity was not yet known. This drivemaximizes interestingness, the first derivative of subjective beauty orcompressibility, that is, the steepness of the learning curve. It motivatesexploring infants, pure mathematicians, composers, artists, dancers, comedians,yourself, and (since 1990) artificial systems.
展开▼
