The short answer is not much, certainly not enough for anyone to worry about.  Furthermore, the temperature change observed is nothing out of the ordinary.

• In the past 120 years since 1900, there has been a 1.2 degree Celsius increase in global temperature (Lindzen, p. 12).
• The warming from 1910 to 1945 was real; it is confirmed by thermometer records as well as proxy data, but it occurred before human greenhouse emissions could have caused it. The warming that may have occurred from 1978 to 1997 is almost entirely fake, an instrumental artifact found only in one heavily manipulated and unreliable database of surface observations (Singer, p. 190).
• There is an important disagreement about the temperature record since 1979, with data from satellites and radiosondes (carried by weather balloons) showing only a light warming of approximately 0.1^o C per decade … whereas surface temperatures show a warming trend about three times as great…… It is important to note that atmospheric data taken with balloon-borne radiosondes independently confirm the satellite data (Singer, pp. 107-108). I believe temperature data time series from satellites are more reliable than temperature measurements taken from the Earth’s surface because the conditions under which satellite measurements are made – outer space – don’t change.
• The variations in temperature from some average breathlessly reported in the mainstream media are small and totally swamped by entirely natural annual variations that humans take for granted and handle with ease. The very slight upward trend in global mean temperature should bother no one.
• There is considerable uncertainty whether the average global temperature reported for a given year is the true number. The variations in data from individual weather stations around a global mean annual temperature are huge.  Scientists call data with very large variations around a mean “noisy”. The more noisy the data, the greater the range of uncertainty as to whether the average reported is the true number. Consequently, fluctuations of a tenth of a degree or two are unlikely to be significant (Lindzen, p. 4).

To put the increase in average temperature in in context, the difference in average temperature from January to July in … major cities ranges from just under ten degrees in Los Angeles to nearly 30 degrees in Chicago. And the average difference between the coldest and the warmest moments each year ranges from about 25 degrees in Miami [45 degrees F] to 55 degrees in Denver (a 99 degree Fahrenheit change) (Lindzen, p. 12). It escapes me why any honest observer would be alarmed by a tiny change over more than a century in the typical annual range of temperature. Set aside for the moment the likelihood of significant sampling errors that would bias temperature measurements up. Even if the temperature record is accepted at face value, no human, plant, or animal is going to notice that they typical annual range of temperature since 1900 has increased from -10 to 40^o C in 1900 to -9 to 41 today.

• When you see a temperature change chart, it is important to pay close attention to scale. When climate alarmists or climate realists talk about the global mean temperature record, what they really are talking about is the variation in temperature around a thirty year average. Scientists call this variation around the mean (deviation from average over some time period) the “global mean surface temperature anomaly”. These variations are measured in tenths of a degree Celsius.  Alarmists can make a little, insignificant change look like a big, meaningful change.