Unfortunately no computer model prediction 100 years into the future has ever been tested. That's the essence of science: propose a theory and then back it up by testing its predictions.

Models are getting good now at predicting the past, i.e., they can correctly predict 1980s climatology from earlier data without depending on too many "free parameters". But of course the experimenter already knows what the "right" answer is in this case. To gain credibility models need to predict, say, 2012 climatology and then have it come true (and do it more than once to avoid luck).

Climate prediction is just a very hard problem. Historic data is untrustworthy (in the case of tropical cyclones "historic" means any data earlier than 1980(!) may be questionable) and nobody knows what cycles dominate over which time scales. The computers are good enough but we just don't know what to put into them.

It would be very nice to be able to plan and prepare for a warmer or colder world in the future but there are no credible (i.e. a history of tested predictions *that came true*) forecasts of that sort.