That'll most definitely happen at some point: https://openai.com/research/improving-mathematical-reasoning...