I agree it's complicated, but are we talking about more than 50 years?

That's not really a question I can answer, but I wouldn't be too surprised if it did.

Basically, to do it right, we'd need a lot of different fields that are in their infancy right now to mature more-or-less all at once. We're talking about taking macroscopic structural issues like brain volumes in different regions or the size of axon paths between them which are vaguely described by modern neuroscience and the huge number of genes potentially implicated in intelligence and then turning that into a single theory of what 'g' actually represents and then using that as a basis for therapy. And that will very likely depend on how good our ability to transfect large amounts of DNA into living cells are; right now you can't even move the gene for hemoglobin, much less whatever is needed for the hundreds of alleles you might have to change in a full overhaul. Throw in 15-20 years for FDA approval and your standard political controversy and you're looking at a big question mark.

I mean sure it could be that the problem isn't as difficult as it looks or the Singularity materializes and drops a solution in our laps, but I really wouldn't advise you to count on it.

Claimed dysgenic effects are very slow.

The problem is that the rate isn't constant; Vining, the guy people usually refer to for the "whites lose 1.6 IQ points a generations" figure, pointed out that the fertility differential is highly dependent on overall population growth. In periods of population increase the high IQ tend to lead the pack and the trend is eugenic, while in periods of population decline they trail and create a dysgenic trend.

Currently the US birth rate is below the death rate, with higher-IQ whites and east asians having the worst of it, so it wouldn't be surprising if by 2048 we see a much larger decrease in global IQs than you could have predicted with data from the 1940s birth cohort.