2 ms·

So it is safe to say the hype from people like Steve Hsu about BGI pioneering vast increases in human intelligence is mostly flash and no fire?

• gwern 9y ago
  Well, it wasn't wrong, necessarily. A large project in 2013 to start uncovering the genetics of intelligence did in fact finally succeed and end the debate over the heritability of intelligence once and for all. It just wasn't BGI's project... Hsu has acknowledged that it didn't work out, but BGI's primarily organizational failure, and intelligence/genetics research since then IMO, doesn't undermine any of his broader claims as expressed in https://arxiv.org/abs/1408.3421 https://arxiv.org/abs/1408.3421 but supports them - intelligence is definitely highly heritable and reasonably close to the classic twin estimates, mostly genetically correlated with good things so boosting it will have mostly unintended good consequences, mostly additive so the extrapolations will work out at least several SDs, etc, all of which implies that very large boosts in intelligence are possible. And if BGI had been able to merely 10x its sample size, it really would have managed to find interesting things due to the enrichment & better phenotyping than SSGAC; but it was too small then (the approach works better, just not better enough). The main implication of the BGI failure is mostly that selection has a bit more potential than thought (more small variants to gradually increase), and editing will be a bit harder than thought (no big rare variants to edit one by one).