People tend to confuse climate and weather, i.e., it's cold out today so there isn't global warming and "we can't predict the weather a week in advance, how can we predict climate years in advance??!11!".
Just to clarify...
If I gave a climate model all the temp data from 1888-1990 and then said ok, what's the AGT in 1995? Would it be correct? I guess the difficult part would be finding a model that hasn't already been modified so it would be correct for 1995. You'd have to use the 1990 version of the modeling software.