PS -- The Arctic Ocillation and its effects are probably independent of the secular global temperature trend. The FT article that Angelus cited seems a little confused -- my understanding is that you can get more snow from warmer temperatures due to warmer air being more able to take up moisture to dump later, and snow reflects more radiation from the sun which might cause local cooling, but they don't make that point in the article.
Doug's point about wilder swings is more on target, except that the growth in the mean is outpacing the growth in the variance, that is, the swings in temperature globally are widening, but the bottom end of the tail is staying put. For the US, there have been more high temperature records than low:
https://www2.ucar.edu/atmosnews/news/1036/record-high-temperatures-far-outpace-record-lows-across-us
For a fairly accessible global view, see here:
http://www.columbia.edu/~jeh1/mailings/2012/20120105_PerceptionsAndDice.pdf
Figure 4, of temperature anomalies (ie differences from 30-year mean), illustrates this nicely. The mean of the anomalies is increasing (indicating average warming), but the distribution is also spreading (greater variance, ie extremes from the decadal mean temperatures). It's just not spreading enough to create more low extremes. I suppose in statistical principle this would allow low extremes to be more extreme, but that's contingent on physical constraints.
-- Andy