I wonder if you could trivially make the model better at math by hacking a precise calculator into its model somehow that it naturally figures out how to use. A

I wonder if you could trivially make the model better at math by hacking a precise calculator into its model somehow that it naturally figures out how to use. And whether you could do the same for human brains.