I do anticipate it, but in the situations I'm asked to do such calculations, I don't usually have the option of refusing, nor would I want to. For most real would situations, it's generally better to arrive at a ballpark solution than to refuse to engage with the problem.
In the very unserious hypothetical I'm describing, I'd say Lloyd's capabilities match that of GPT-4. In this case, he's not a calculator, but he is a decent programmer, so like GPT-4 he quickly runs the operation through a script, rather than trying to figure it out in his head.
