100 years is not very long-term at all. The added CO2 will not miraculously disappear after a century, but will go on warming the planet. (And that's just assuming that humans stop burning fossil fuels now.) There are all sorts of feedback effects that will come into play at some point, if they haven't already.
For one example, a lot of CO2 winds up dissolved in the oceans, and the solubility is inversely proportional to water temperature. As the ocean warms due to human-generated CO2 in the atmosphere, it becomes less able to serve as a CO2 sink, and eventually would release some of its existing CO2, causing further warming. The ocean mixing time (between surface and deep ocean) is on the order of 1000 years, so whatever is set in motion today keeps on happening. The warming ocean might also trigger the release of methane clathrates: https://en.wikipedia.org/wiki/Clathrate_gun_hypothesis
However, we don't really have to theorize that much about the long-term effects, since we have a reasonably close parallel from Earth's geologic past: the Permian-Triassic extinction event: https://en.wikipedia.org/wiki/Permian%E2%80%93Triassic_extinction_event One leading theory for the cause of this is that major vocanic eruptions (the Siberian Traps) took place in an area of coal beds, setting them on fire. This of course added lots of CO2 to the air, just as human use of fossil fuels has.
So the hypothesis is that the current warming (and future warming that is already "baked in" by the already-released CO2) will trigger unstoppable feedback effects. Not quite a runaway "Venus Effect" (that's probably impossible on Earth), but one that causes mass extinctions.
Some people think that humans would survive those mass extinctions. I doubt that*, myself. Too many people live lives divorced from the natural world (thinking for instance that food comes from supermarkets & fast-food joints), and so fail to appreciate just how strongly human survival depends on the existence of the global ecosystem.
*Of course this is opinion. The only way to test it (other than computer models) is to perform the experiment, which is analogous to checking whether a gun is loaded by pointing it at your kid and pulling the trigger :-)