CO2 is transparent to visible light: E.g., pop open a brewski and observe that can't see the CO2 and that it does not cast a shadow.
CO2 absorbs photons in three narrow wavelength bands, one for each of bending, stretching, and twisting of the molecule, out in the infrared. The absorption spectrum is shown at the Web site of the US NIST.
That absorption makes CO2 a greenhouse gas.
The usual claim is that the infrared does not come from "sunlight" but from Planck black body radiation from the surface of the earth.
The broad, intuitive idea that the CO2 acts like a thick insulating blanket over all the earth and trapping all heat is not very close to reality.
Then there is the question of, how much warming does CO2 cause? Some dozens of studies, with computer models, etc. were done. Results varied. A summary of the results is in a graph at
http://www.energyadvocate.com/gc1.jpg
Also it appears that the names of the studies are in small print in a box in the upper left of the image.
Some results predicted significant warming, but the times of the predicted warming came and went years ago without the predicted warming.
Al Gore's movie showed a graph of data from ice cores from Vostok station in Antarctica. The ice cores went back
800,000 years of so. Gore graphed both temperature and CO2 concentration and observed that the two went up and down together.
As I read the graphs:
(1) When temperature started to rise, CO2 concentration was low, not high. So, something made temperature rise, but it was not high CO2.
(2) About 800 years after temperature rose, CO2 concentration rose, and the usual explanation is that the extra CO2 was from more biological activity from the higher temperature.
(3) Some thousands of years later, temperature fell. Then CO2 concentration was high, not low. So, something made temperature fall, but it wasn't low CO2 concentration.
(4) We had some CO2 from WWII and the post war economic boom, but from 1945 to 1970 we had some significant cooling.
For one more, not from Gore's graph, recently I heard a claim that the three highest temperatures on record happened in just the last few years or some such. There is some issue about the relevance of extreme values: E.g., in any stochastic process, of course the highest values observed so far never go down, and in real processes, say, the fastest 100 yard dash, we can expect the highest values to go up.