Due to the extreme data quantity requirement for pre-training, LLM effectively locks your reasoning language into the lowest common denominator, aka Python. Sur

Due to the extreme data quantity requirement for pre-training, LLM effectively locks your reasoning language into the lowest common denominator, aka Python. Sure, people (maybe very smart) can come up with reasonable, effective, efficient notation; the problem is to train the model to use it properly.