The question for them is similarly existential - who is going to pay for research grants when the machines do all the work? Why pay for a PhD student to work on some research topic if an llm already proved the theorem?