Making computers to learn and make decisions is done at present especially in clinical conditions where the parameters used for diagnosis are vague. For example when poorly defined parameters are used in diagnosis in radiographic examination of tibia fracture healing. In recent past digital data in computers had made processing of data both simple and cheap. Data like electrical output from the fractured limb can be fed and a computer can be trained and then later tested for a new fracture case of unknown nature. This can give precise idea of the exact day of healing of the new fracture. This exploits the sharpness of the diagnosis with computers i.e using artificial intelligence , even with some lack of transparency. The main limitation is the lack of confidence on the clinician side. But the potential of artificial intelligence is more than it is used nowadays.