Vladimir Vapnik, co-inventor of Support Vector Machines and professor at Columbia University, discusses the limitations of current artificial intelligence approaches and the open problems in the field, as well as his thoughts on the nature of learning and the question of whether God plays dice.