AR(2) forecasting problem

[olpa]

I would kindly ask if someone could help me on this... I have 200 observations on which I applied a AR(2) model : X_t=a1X_(t-1)+a2X_(t-2) +e_t using the first 190 observations of them on eviews.

The question I have to answer is "assuming that a1, a2 are the real parameters (not estimated) of the model and knowing only the first 190 observations, firstly I have to forecast the ten last ones (X_191 to X_200) and then to give on a list the a1^h and a2^h for h=1,...,10 where X_(T+h)(estimated)=a1^h*X_T + a2^h*Xt-1.

Does anyone have any idea what may this means? I made the forecast of the last ten observations, but why to have a1^h and a2^h?