The concept of probably approximately correct (PAC) learning has been foundational in computational learning theory.
I'm not perfect, but I'm always striving to be better.