This one is interesting.

It seems to me like you can set up the lagrange interpolating polynomial and just hand it any n primes as its first n integer values, and you'll have the desired result.

Perhaps the idea is to find a polynomial with only integer coefficients, because that makes it significantly harder, since the coefficients of the interpolating polynomial are likely to be non-integer rationals.

This is an interesting puzzle..