Agreed on all parts. What is being said makes sense on the surface but I am not taking it as gospel until there is direct causation confirmed.
I'm not being a dick, but what confirmation of direct causation would be enough for you? I'm just trying to think of how they would even track it down specifically.
Right now, they are seeing a rise in CO2 levels in the atmosphere that corresponds with our industrialization, and a corresponding increase in the rate of warming. It's not like we can conduct experiments: "OK, let's release 1M cubic meters of CO2 in the atmosphere. By our calculations the CO2 levels will increase by .01% which should cause mean global temperature to increase by .05 degrees C"
Likewise, we can't reduce CO2 levels in the atmosphere unless we plant a bunch more trees or stop burning so much fossil fuel. If we could then maybe we would see that yes, CO2 levels did come down and so did the mean global temperature.
So, that's where the models come in. I don't want to say that the models will NEVER be accurate, but they'll never be accurate. LOL. That's just how models of complex systems are. I worked with somebody that was working on a linear 6-DOF model using computational fluid dynamics to accurately model the lift of helicopter rotors. However, depending on what the model will ultimately be used for, you could possibly just do a simple model of X rpms = Y lift. Is the second model "accurate"? Yes and no. It's not 100% accurate because it won't be sophisticated enough to take into account every environmental factor, but it's accurate enough for it's purpose, i.e. this much throttle means this change in RPMs means this much change in lift, equals this much gain in altitude. That's "good enough" potentially, for a flight simulator. With the 6-DOF model you can do what-ifs regarding blade design changes, that still won't ever be 100% accurate.
Now you apply that same reasoning to climate models. Helicopter rotors are well understood, and relatively simple in comparison to the entire planet and everything on it and everything external affecting it. You could have the model absolutely 100% correct mathematically, but it still won't be accurate as a predictive model because of random events. Let's say China opens 100 more coal powered power plants, you didn't account for that when you made the prediction, so your prediction is now wrong.
I'm going a long way to say that I'm not sure what you're looking for is even possible for this particular subject.