Jeff Masters tells us that the modeling work of Dr Peter Stone (MIT) suggests that current predictions of a rise in global average temperature of 2.5-10.4 F by 2100 may, in fact, underestimate the actual increase in mean global temperature.
Jeff Master's layman's summary
The primary distinction, based on Jeff's summary, is that in the older models, the oceans were expected to be better heat sinks than they appear to be functioning as today.
As a point of reference, information from the EPA indicates that the mean global temperature has increased approximately 0.5-1F in the last century.
EPA numbers
That relatively modest (in comparison) temperature rise already has given rise to us talking about shipping lane disputes in the Northwest Passage, hurricane seasons that dwarf last summer's, and rising sea levels that are likely to endanger large fractions of the world's population.
Regardless of what one's personal politics might be, it's past critical that we attempt to understand how our world is likely to change and how we might be able to mitigate or cope with that change. Barring a Krakatoa-esque eruption, we're talking about potentially disastrous changes that likely will happen in our lifetimes, not on geologic timescales.