I don't think this is equivalent. The calculator won't occasionally hallucinate a wrong answer. LLMs are a far leakier abstraction which means skill atrophy impacts evaluation and verification ability.