The IPCC 'projections' for temperature rise over this century, given in Table SPM3 on page 13 of the SPM, range up to a scary 6.4 degrees. The associated text says "Assessed upper ranges for temperature projections are larger than in the TAR (see Table SPM.3) mainly because the broader range of models now available suggests stronger climate-carbon cycle feedbacks." Such a projection seems far-fetched, given the 20th Century temperature rise of around 0.6 degrees, and the established logarithmic effect of carbon dioxide, meaning that more carbon dioxide has less effect.
So where do these unlikely projections come from? Page 12 claims that "Water vapour changes represent the largest feedback affecting climate sensitivity and are now better understood than in the TAR. Cloud feedbacks remain the largest source of uncertainty". This is based on computer modeled “positive feedbacks” from increased water vapor and clouds, specifically designed to increase the impact of anthropogenic greenhouse gases on temperatures by several times the values supported by theoretical physics or by the actual 20th century temperature record (which provides no evidence of such “positive feedbacks”).
Part of the argument for this feedback is that warmer temperatures lead to more evaporation and hence more water vapour, another greenhouse gas. But we have already seen that there is no increase in water vapour.
What does the IPCC mean by "Cloud feedbacks remain the largest source of uncertainty"? Here it is necessary to interpret the IPCC code and read between the lines, or look up the relevant section of the main report, section 126.96.36.199 on p. 635. Here it is indirectly acknowledged that cloud feedback is negative, since warming leads to more evaporation, more cloud formation and hence more reflection of sunlight. Another mechanism whereby water vapour feedback is negative was discussed by Lindzen (1990) Some coolness concerning global warming, Bulletin of the American Meteorological Society, 71: 288-299.
In general, in nature, negative feedbacks are far more common than positive ones. Negative feedback gives stability of a system to external perturbations, while positive feedback leads to instability and exponential growth. If there was a positive feedback mechanism between carbon dioxide, temperature and water vapour, as hypothesised by the IPCC, it would have led to large fluctuations in the past.