While small changes in atmospheric CO2 levels do correlate with recent ocean surface temperature changes, broad trends in CO2 levels correlate strongly with average atmospheric temperature. This is derived from ice core data going back about 420,000 years.
The ice core data shoes a correlation that is not linear, but rather, matches the solubility curve for CO2 and water, with respect to temperature, but the best fit is obtained by introducing a time lag between the temperature and CO2 level data points.
The CO2 levels change a bit more than a thousand years after temperatures changes. Thus, the current (last couple of centuries) CO2 increases are actually caused mostly by the Midieval Warm Period rather than anything that is happening now.
One could go through a hand waving exercise to explain why there would be a 1000 year time lag, but the important point here is that the time lag exists, not why it exists.
You are absolutely correct. The difficulty in discussing AGW is due the layer upon layer of fraudulent science and positions the global warming fear mongers have built their case on. The ice core sample fraud is a whooper by any stretch.
along those lines I give you this from the above link "The problem in saying humans caused a 30% increase in atmospheric CO2 is that the earlier measurements were made in ice core samples, while present measurements are made in the air. The different methods of measurement are not comparable, because the ice core measurements are extremely dubious. The method was to crush the ice sample in a vacuum and "rapidly" measure the contents. The measurement is a concentration relative to the nitrogen and oxygen. Everything about such a procedure sends up red flags. How perfect can the vacuum be without destroying the ice sample? A vacuum cannot be created rapidly; so how could rapid crushing be relevant? A vacuum would explode the gasses in the ice causing them to mix with the gasses being evacuated. Then there's a problem with the sample. How stable can the components be over time? There would be a high tendency for CO2 to react or migrate, which would reduce the peaks. Nothing about such a procedure points to an absolute value which can be compared to other methods of measurement. All indications are that measurements in ice are much lower than real values. There were direct measurements of CO2 in the air back then, and they show much higher values than the ice core measurements...
http://www.21stcenturysciencetech.com/Articles%202007/True_CO2_Record.pdf