That's where new species arise, but not what they are. By the dawn of the age of exploration, the human race had spread around the globe. The varying environments and the isolation of various subgroups from the rest of the gene pool had produced the beginnings of speciation processes that would eventually have gone to completion had we not re-discovered each other and re-connected all the gene pools. Now we're basically re-melding.
Thus, up to a somewhat blurry point, the process is reversible. Put in barriers to gene-mixing, differences arise. Take out the barriers, re-mixing can occur. (But in nature, it doesn't always. Sometimes the process once started simply runs away because of sexual selection pressures or a lack of situational viability of hybrid types.)
... or hybrids of the type similar to when horses and donkeys mate to produce mules.
The product (a mule) of horse-donkey hybridization is not a new species. It's not even fertile. (It's a useful farm animal combining some of the better points and skipping some of the drawbacks of its parent species.) You misinterpret the significance of the situation. That horses and donkeys are cross-fertile but produce sterile offspring is a sign that speciation has already occurred and relatively recently.
Despite all the similarities between the species and the near-compatibility that allows the production of mules, there is no way for horse and donkey populations to re-meld in the way humans are doing. They are on the other side of the speciation barrier and will go their separate ways now.
We still have populations that seem quite isolated. Have there been any studies about whether the Congo pygmies or the Kalahari bushmen mingle reproductively with the surrounding populations? The same question applies to other isolated groups like Australian natives, Amazonian tribes, and highlanders in remote Indonesian areas. Even if there were occasional "melding events," it may be that the bulk of such populations continue to be reproductively isolated, and if this continues, it's conceivable that speciation could result in a few tens of thousand of years.