If the basic program allows the machine to gather new knowledge and interpret it on its own, outside of the constraints of the original programming, then it has become AI. Until then, no true personality or being has come into play. ;P
However, if we're talking hypothetical, simply learning does not mean feelings or emotions. If you start with strictly minimal facilities, there is no need for emotions for decision making. There would be no instincts, no drive, just an electronic "brain". The stories always go on about how robots learn to "feel" but why? What's the need? Why is it necessary?