Isn't it a little too fast for "evolution"?
Not really?
If there's a range of "how aggressive a bear can be", and it's mostly driven by genetics, and aggression is heavily selected against in the environment? Then you can get a considerable reduction in aggression in the span of as little as a few generations. Bear generation time is what, 5 years? They coexisted with humans for a long time now.
Now, traits with weaker genetic components (i.e. if bear aggression is only 50% genetic) can take much longer. Even more so for traits with low variance, or highly complex traits and behaviors. But evolution isn't always slow. Certain changes can happen quickly - about as quickly as you can apply the selection pressure.
Evolution works in bursts. Species can stay stable for millions of years and then evolve in relative blink of an eye when the environment changes.
Right, how do you know the gene pool now mostly contains large aggressive bears that instinctively stay away from villages, and small cuddly bears that are enjoying left over pasta suppers ?
Maybe it's just that many of the large aggressive bears living near villages have just been shot or scared away, but the genetics is unchanged and the offspring of large aggressive bears currently living away from villages will have no aversion to trying their luck in the village ?