For starters, if it were superintelligent it would eventually make discoveries. New discoveries were not in the training set originally. The model needs to be t

For starters, if it were superintelligent it would eventually make discoveries. New discoveries were not in the training set originally. The model needs to be trained to use the new discovery to aid it in the future. As it is, it has to keep "rediscovering" the same thing each and every time, no matter how many inferences you run.