If you think #LLM are useless because they can't do the math, know that you are in fact allowed to do this:
@faz don't solve the problem, write the code to solve the problem brings it back down to tractable "language autocomplete probabilistic matching". powerful technique!