There has been some criticism of the 20 global climate models (GCMs) relied upon by the IPCC, specifically that they deal poorly with clouds. All 20 assume a positive feedback from warming - the warmer it gets, the better the greenhouse effect works. (Simplified, the greenhouse concept is that visible light is reflected off the planet, and the infrared heat is captured by greenhouse gases, including clouds, water vapor and to a much lesser extent since it is a trace gas, CO2. Positive feedback means that emissions into space should therefore decrease with warming.) The IPCC admits it handles clouds poorly; if warming leads to more cloud formation, light is reflected before it ever gets to the planet, and the world cools.
Now a study for other purposes has shown a devastating flaw in all 20 models: where they predict lower emissions, recent satellite measurements show higher emissions with warming, sometimes as much as 7x the prediction. See this chart, where the red line is the measured emissions in space, and the dotted line is the mean prediction of the models. (The gray is the band of predictions around that mean.)
Bottom line: where the models predict a positive feedback (more warming leads to more greenhouse absorption of infrared), the satellite measurements show negative feedback - more warming leads to less greenhouse effect. The climate sensitivity in the models to CO2 is much less than predicted. As Dr Lindzen concludes (emphasis added):