But the point is that you can solve these problems, just that it takes significantly longer to solve them than it does to check the answer once you have it.

So you decide to do a brute force calculation on a computer for three days, write down the answer and tell her (a la kavitra's comment) then she sees the answer on the paper in the future and you change your mind and just check the answer.

Because Alice sees without the prediction, there's no need to have a stable causal loop like pengvado discusses.