4 ms·

Thanks for the correction. Do you think we're currently equipped to breed a super race?

• gwern 9y ago
  No. State of the art embryo selection is something like +0.5 IQ points, which might be profitable but can only be charitably described as 'subtle'. Regular embryo selection is currently too weak because you have only a few eggs to choose from, most of which will not lead to a live birth, so even with the great GWAS results coming over the next few years, the gains will be small. Similarly, CRISPR is not that great because while we have identified lots of SNPs which allow prediction of intelligence, few of them are causal, so you would need hundreds of successful edits to get a truly noticeable boost like 10 IQ points. What is necessary for large gains like +10 is one of: ability to create scores or hundreds of eggs to do embryo selection allowing picking best embryo out of hundreds rather than out of 5 or less (perhaps using egg->stem-cell->egg induced pluripotency & specialization techniques, currently doable in mice/rats but not humans); iterated embryo synthesis using the same induced pluripotency to coax a set of stem cells through many generations of sperm/egg->embryo->selection->egg/sperm for a cumulative gain; CRISPR scaling to hundreds of edits plus better identification of causal variants; and whole genome synthesis to create a single hyper-optimized genome from scratch without bothering with selection/editing. The timelines I would hazard guesses at something like 5-10/10/5?/15 years respectively. So embryo selection right now is a curiosity, but with the right breakthrough, it could become important, and even if it doesn't, at some point CRISPR and genome synthesis will allow for big gains.