I do believe them because many sources from different proxies agree on the temperature record, the science is very well developed and a strong consensus across a very large majority of experts and separate disciplines has been achieved.
"Highly dubious" is how I would describe the fringe criticisms of this work.
Not for the recent period where we have actual accurate temperature data to compare to. Some proxies show warming for that recent period (but not quite the same as the actual temperature data), others (such as the "hockey stick" reconstructions) actually show cooling. The fact that they agree better where there is no actual temperature data to compare them do does not fix this problem.
> a strong consensus
Consensus means nothing in science. Science works by making accurate predictions. Climate science cannot make accurate predictions of future climate. Even the IPCC admitted this (finally) in the AR5 (and then proceeded to obfuscate it by basing its conclusions on "expert judgment" instead).
For example, data suggesting that major storms were more frequent in the 19th century than the 20th.
> World temperatures rose at a rate unprecedented for millennia during the 20th C
Only if you believe highly dubious temperature reconstructions, including their claims to be accurate to within fractions of a degree.