Ahh, you are using Cowen's autocorrelation trend calculator which is why you get the numbers you have.
The autocorrelation method he uses causes there to be a large error range which is why the calculated error is so large.
The 0.05 +/- 0.241 per decade means that the trend could actually be anywhere from -0.191 to 0.291. Therefore, the result using that method is statistically speaking saying the temperature changes (if any) are indistinguishable from a hiatus.
Using a better method and the actual data from the UAH or RSS website and calculating the least-squares linear regression, you get the numbers I have.