Designer babies are coming. These aren't the far future, they'll start in a few years. Probably within a century you'll be able to change the genome of any living person in-place, then designer babies will be quaint. And even during the century where designer babies are the hot thing, I suspect designer babies won't be more than about 2% of the population because it's more fun to just have sex.
There are several technologies that have to work, be legal, and be cheap to get designer babies. They are:
What this buys you is the ability to edit any number of genes for sure. It also allows the embryo to develop without being sampled; the tests of the genes were done before conception not after. It does not tell you the sex. It's like you edited the parents, then the edited parents do the shuffle of sex the normal way. If you want to guarantee the baby gets a particular allele, you have to make sure the parent cell had all copies of that the same.
An alternative already working today is to harvest many eggs, fertilize them all, quickly test all of them to see which came out with the combination of genes you want, then implant that one. That's worse. Suppose you want to choose 20 genes from among the parents. Embryo selection would need on average 220 embryos to choose from, which isn't really feasible, but the CRISPR approach could do it no trouble. If you wanted an allele not present in either parent, that's impossible with embryo selection. Choosing with embryo selection requires looking at every embryo, but choosing with CRISPR means selecting the genes you want edited. Embryo selection requires sampling from the all the embryos after conception, which is pretty severe interference at a very early age and has to be done very quickly. It also requires destroying all but the chosen embryo. Destroying all but the chosen embryo is a big deal if you believe that a full human soul is attached to each embryo at conception, but not so big if you believe an embryo consisting of 16 precursor cells arranged in a sphere is as it looks and acts. The CRISPR approach does not require destroying any embryo ... it destroys skin or stem cells before eggs or sperm are produced.
I just read She Has Her Mother's Laugh, which was talking about this, using CRISPR and producing gametes from skin cells. The book seemed uncertain how the world would react to CRISPR editing human babies: whether parents had the right to shape their children, whether this would reduce human genetic diversity, or what. I know my opinions.
I think parents absolutely have the right to shape their children at birth. Parent's ownership of their children is well approximated by 100% at conception and 0% at age 19 and linear with age in between. 50% ownership at age 9 for example. Parents get those rights as a consequence of their responsibilities: they have to fully provide for the child until about age 19. Governments, keep your fingers out of this please, all rights to choose about the baby before conception and for months afterward belong to the parents.
If designer babies the CRISPR way become easy, what will result?
The result is that for the couple dozen traits the parents can get themselves to care about, they can choose them directly. Humans have 23000 genes, so a couple dozen is probably fairly superficial. The unintelligible remainder would get chosen randomly, probably by normal sex.
What this means for evolution is that known good/bad alleles are heavily selected for/against, more heavily than natural selection or even selective breeding could accomplish. If things go too far, parents could select the originals back, so I don't see any harm in this. And the unknown remainder are mixed randomly, and sometimes you get lucky and sometimes you don't. This is a fantastic way to improve the species while maintaining diversity. I think this is a very good thing: the randomness of sex can be reserved for questions we don't know the answer to (genes that surely matter but we don't know how) rather than questions we already know the answer to (like is it good to have cystic fibrosis).
Homosexual couples would be able to produce their own children. Two males could produce children of either gender, but they'd have to enlist a woman to get pregnant with the embryo. Two females could only produce a female directly, but if they borrow just the Y chromosome from a male they could produce a male as well.
Since the highly valued traits will spread as a matter of culture rather than heredity, but the unvalued genes still spread by heredity, I predict Africa will rise triumphant. Perhaps there are good genes that are universal outside Africa but rare within it (due to founder effects), but those will be few, and Africans can select those by CRISPR. For the remainder, they've got far more inherent diversity than the rest of the world, so their random highs will be higher and random lows will be lower. And it's the random highs that matter.
It is possible some country will bring control freakery to a new high, require all women to have their tubes tied shortly after birth, and allow pregnancy only through embryo transfer. The country might not even allow the parents to produce their own children, they would only allow children bred from a select breeding population. This sounds wildly unpopular to me and liable to produce severe inbreeding.
A nanny state (a state dedicated to enabling the people to live full, reasonably unrestricted, and healthy lives) that requires tubes tied and all pregnancies through embryo transfer is interesting. Unwanted pregnancies from rape or otherwise are automatically impossible. Nobody needs to take birth control or even worry about it. The state could easily not allow pregnancies until after high school. Everyone should be allowed 2 kids if they want them, then the right to additional children could be doled out by the market, or by lottery, or by getting a degree in parenthood, etc. The state would require that all fatal or crippling alleles be edited out (on the grounds that it has to pay for them plus the children would be clearly unhealthy otherwise), but everything else is up to the parents, because the goal is to enable parents and children to live fully healthy unrestricted lives (to the extent possible). The biggest work for nanny states is keeping some citizens from oppressing other citizens.
On a simpler note, CRISPR (without mucking with skin cells and eggs and such) can be used to edit plants. If you figure out the genes that make a food crop, you can apply them all to the wild progenitor of that crop and get the food crop again but with more hidden variety. And if you do it a few dozen times to the same crop, you get a population of that food crop that will breed true and with more hybrid vigour than any of our existing crops. Goodbye Monsanto, at least with the business model of selling the same hybrid seeds year after year.
Even more practical: all these species going extinct, and food crops where the wild progenitors are endangered? In addition to trying to conserve endangered species, fully sequence all the individuals in them that you can find. It might be hard to keep wild cacao trees growing, but it's easy to keep a few thousand of their genomes preserved on hard drives once they've been sequenced, and CRISPR will let us easily find and use all those variations forever. Or take and freeze samples now and sequence them later.
I have a rosy outlook on gene editing in humans and animals and plants. (Am I saying I support eugenic? Absolutely yes! But only in the case of parents choosing their own children.) I'm quite scared of gene editing in viruses and bacteria. The worst that happens in gene editing in humans is, you quarantine the human that comes out badly. The worst that happens in gene editing of bacteria and viruses is that everyone quickly dies. And it's pretty easy to reach that worst-case outcome. Normally I'm in favor of all information being free. But DNA sequences for designer viruses on the web (coupled with equipment angry teenagers can use to produce them in their garage) are a strong argument for at least some forms of information being forcefully censored. Why, for example, is the smallpox virus sequence public information?