I see no justification for that kind of prognosis. We do not even have a mathematical theory about how these neural networks learn, let alone a theory about how they might reason. Super AI is super speculation at this point.