do you think that would be the case if you had a supercomputer with billions of GPU clusters trained on every single digital data ever generated by the people involved in the study?