This is a critical review of recent results in stochastic approximation methods for optimization problems. Part I deals with unconstrained optimization, with special emphasis on the problem of choice of the direction and the length of iteration steps.