A recent scientific study showed how, hundreds of millions of years ago, there was a land-locked ocean over the North Pole. That lead to a massive decline in atmospheric CO2. That is nothing like the current situation, but it does show how feedbacks of warming can lead to reducing CO2 levels. Such negative feedbacks are likely to overshoot. For example if a warm ice-free Arctic does lead to something that reduces CO2 levels then it is likely to continue to do so for much longer than we would wish.
Another reason why I think global warming could lead to cooling is the history of recent inter-glacials. They have been warmer than this one, but have not lead to runaway warming when the Arctic melts. Instead each one spikes and then crashes:
Here's my graph of the global warming position:
I also reckon that it is simplistic to imagine this 2d picture. Suppose we imagine this 2d graph embedded in a bigger 3d graph. We might role over the top to the right, but then get caught in a channel that leads back around to that waiting ice age.
In summary I want to see more science before we do any geo-engineering: including attempts to get CO2 levels all the way back to the "natural" 280ppm.