This soundbite does not carry much useful information. Sure, it is an amusing factoid that the spreads of the data points overlap, but that does not mean anything, especially if you do not explain what that spread represents (is this a confidence interval for the true average, or is it just meant to represent the spread (due to geography) of the data over which it was averaged). It is completely nonsensical to use this margin in order to say things like "Year X was as hot as N other years".

Moreover, even if the margin of error is big, we can still easily calculate (with high precision and certainty) the presence of a slope when analyzing the whole data set.

P.S. A minor side note is that this is temperature averaged over the entire globe, so to a big extent this spread is not an "error", rather it is representative of the variance of Earth's temperature at different geographical locations and at different times throughout the year.