Harvard University mathematician Melanie Wood stated that nobody understands an AI model's solution at the time of release