Just thousands, to be precise the last two thousand.
We know for a cast iron fact that this is false. Human populations show the signs of natural selection for resistance to various diseases.
The population increase that humans went through in the last 2-3 thousand years has few parallels in nature, and there is plenty of evidence showing that selection does not operate in expanding populations, I can dig up refs if you want. But to put it simply, natural selection operates under the premise that some of your progeny will die (and with it presumably your "weak" genes), so in order for natural selection to operate, a certain percentage of the population has to perish. But in today's society this is not happening anymore, or in other words, both the fittest and the weakest are surviving. If both are surviving there is no selection.
I would be interested in your refs because the description you have given makes no sense. All selection needs to act is inheritable differences in relative fitness - a gene that produces more living offspring can be selected for in an expanding population as easily as one that produces a drop in living offspring in a reducing population. It would only make no difference if both the fittest and the weakest survive unless they both also go on to have exactly the same number of surviving offspring themselves - at which point they have the same fitness anyway.
It's also simply not true that everyone was surviving for last two thousand years, throughout most of that period the majority of people born never lived long enough to have children of their own.