In an interview, Pinker says,
I’m skeptical about that we’re going to see enhancements of human nature by genetic engineering, nanotechnology, or neural implants (though these technologies may be used to mitigate disabilities, a different matter). We now know that there is no “gene for musical talent” that ambitious parents will implant into their unborn children—psychological traits are distributed across thousands of genes, each with a teensy effect, and many with deleterious side effects (such as a gene that makes you a bit smarter while increasing your chance of getting cancer). Also, people are risk-averse (sometimes pathologically so) when it comes to their children and when it comes to genetic engineering—they don’t accept genetically modified tomatoes, let alone babies.
Just before I read this, I posted the following on a private discussion forum:
For those of you have read The Diamond Age, what feature of the future Stephenson depicts there do you find least plausible? I’ll nominate the Illustrated Primer. I bet that no educational technology that relies on communication with the student will ever prove as successful as the primer is portrayed. When it comes to achieving dramatic gains in cognitive skills, some form of biological intervention will prove workable sooner.
When I was in high school, SAT tutors were unheard of. The whole concept would have seemed distasteful. What parent would be so neurotic and competitive as to get their kid a tutor for the SATs? But once a few parents started doing it, other parents thought that they had to do it in order to keep up. Nowadays, I get the sense that any affluent parent who does not get their kid a tutor feels like they are handicapping their child. I’ve been predicting that in another generation, biological enhancement will go through a similar phase change–going from unthinkable to commonplace very quickly.
In your comments, please address substantive issues, leaving out your personal opinions of Pinker or me.
I found any kind of SAT prep distasteful in high school. While I still do, I probably would make my kids do it anyway.
You don’t need a tutor. You need 10-25 hours alone with an SAT prep book. Doing practice tests being the most important part of that.
That initial commitment brings about 90% of the rather modest gains you can make on the SAT through effort. Its all diminishing returns after that. Some exceptions can be made if English is your second language, in those cases you can sometimes better your verbal score substantially with extra work.
Notice his exception for “mitigating disabilities.” Framing matters. In a ‘curing disability’ framing, it’s not so much that people will be ‘adding’ positive genes for ‘enhancements’, but substituting out those single-or-few nucleotide variations that are ‘holding Johnny back’. Consider the case of genetic load of deleterious mutations. It’s not so much that complex features usually judged to be positive are ’caused’ by having a very different kind of gene than everyone else, but instead by having less genetic load accumulated among the many genes that contribute to the expression of that attribute or capability. Selection pressure is currently very low, and lots of people are carrying around a lot of bad genetic load.
So, “Removing / Correcting deleterious genetic load,” will not only have a good socially acceptable framing, but just as with ‘curing disabilities’, I think one can appreciate right away how such interventions would become a matter of entitlement to provision of genetic remediation services from public health agencies or providers, a soft ‘right’, that would also help reduce inequality and the unfair privileges of the rich. Identifying instances
The question is, how much an improvement are we talking about in the average person if we remove most of his genetic load? My read of things is that this improvement could very well be shockingly large, with significant and unpredictable consequences for society.
A couple of points.
1) Embryo selection for specific genes is already common place, you can get an embryos full DNA sequenced. What’s missing is statistical predication for complex traits we care about. This is coming. It doesn’t have to work great, if it works enough on the margin it’ll make a difference especially after a few generations. Look up IVF with PGD.
2) IVF is much harder than people realize, and only about 40% effective for any given cycle. That number might be low because most people trying IVF currently have fertility problems. But most IVF cycles end in miscarriage. That is a hard thing to go through multiple times. Plus the shots that the women have to take are really painful. They go in the same place every day for months, and by the end every shot leaves you crying.
3. IVF isn’t cheap.
Statistical prediction of complex traits will happen in the next 5 years. It’s already here to some degree. But making IVF more successful, less painful, and cheaper, that might take longer. I think those things are bigger obstacles than anything else. I don’t doubt the culture can change to support an expectation of going through a difficult process, but that will only happen after the successes become too hard to ignore.
The biggest problem with genetic enhancement of intelligence is the fact that humans are already very optimized for intelligence via evolution. We have brains so big we have trouble giving birth, which is very rare among species. Giving birth is evolutionary important!
The trade off we want to make is higher intelligence in exchange for requiring more food, because food is cheap now. But I suspect we won’t be able to make that trade off directly by using correlational methods of genetic manipulation. Intelligence is correlated with a host of cognitive abnormalities like schizophrenia and autism, and we don’t know nearly enough about how the brain works on fundamental level to know what is causal or how to get the good stuff while avoiding the bad.
To make it worse, there is going to be an awful lot of variation in intelligence and cognitive stability even among people who are genetically very similar. This means that there won’t be the kind of fine grained control on how we make these trade offs, say by increasing intelligence in exchange for high functioning autism. It would be +5 to +20 IQ points in exchange for a 50% chance of high functioning autism and a 30% chance of totally debilitating autism. Parents are not going to want to take that risk.
Violating your proscription, let me say that I consider both you and Pinker to be valuable assets to our culture. On the issue of genetic enhancement I think you are right, at least in the long run. The complexity of the genome will make such enhancement difficult, but it is potentially very rewarding. The pressure for it will in the end prove irresistable.
Steve Hsu does the best public blog discussion on this I’ve seen. Some of his conclusions are already noted above.
I think the moral arguments against it will have zero restraining power over even the short/medium term. However, as others noted uncertainty, hardship, and cost will be limiting factors in at least the short term. How many X IQ points is worth an Y% increase chance of a bad trait? Even with the data sets that will exist in five years, do we really feel confident in the X and Y% any IVF clinic will give us? How long does it take to get enough data to feel confident in our guesses, does the first generation its done on have to grow up?
I have no doubt that when there are good enough answers to those questions people will be lined up out the door, but its hard to know if that is a short, medium, long, or impossible term solution.
I think of genetic engineering (or just IVF embryo selection) like flying cars. Its fun to theorize about, you hope it happens, and I support a flying car in every garage. However, if someone proposed ending all road maintenance today because flying cars will make roads obsolete in five years you would think that was putting all your eggs in one dangerous and unpredictable basket.
I’m inclined to agree with Pinker on this one. I see genetic modification as being akin to government interventions in the economy. We’d be making adjustments to such a chaotic, stateful, interdependent system with (in my view, hopelessly) overly simplistic models.
Even if you get some degree of mappings between genes and phenotypes, there’s the whole other question of what even makes sense to optimize for. In the language of process control, what is your “objective function”?
There’s also the question of feedback loops: it takes a long time for a human to mature to the point where they are capable of reproduction, and likely much longer to find out whether the modifications you made results in the outcomes you are looking for (assuming you’ve even established what those outcomes are). Trying to shorten the feedback cycle is problematic because it affects the outcomes. Consider the traits that we may consider to be most desirable (e.g. intelligence), to achieve those traits ultimately required lengthening the feedback loop (through an extended maturation process).
Part of the danger comes in too quickly settling on a genetic modification strategy that seems good in the short term, but is disastrous in the long-term. In the language of “explore vs. exploit”, it leads to premature exploitation of a strategy.
In nature asexual reproduction is a form of reproduction which favors “exploit” over “explore”. It enables an individual to quickly dominate a niche through producing offspring, but is very vulnerable to changes in the environment. Hence, the value of sexual reproduction.
Indeed, sexual reproduction, and specifically the process of sexual selection, is indeed a form of genetic engineering. One that I think has a lot of advantages over direct genetic manipulation:
– It operates at the right level of analysis: phenotypes instead of genes
– it strikes the right balance of “explore” vs. “exploit” in terms finding optimal genetic configuraitons
– It is highly decentralized: it requires considerably less capital investment
What I’m trying to say is that I think making babies the old-fashioned way is in general going to be superior to attempts to “plan” our biological future at a genetic level. This is not to say that a population that engages in genetic modification wouldn’t be able to overcome a population that refrains, but that a population that prefers “old-fashioned” reproductive strategies is going be able to better survive a changing environment than one that does not.
IVF went from scary “test-tube babies!” to matter of course for 40 year old women trying to have kids. If there’s a successful genetic intervention, it won’t take more than a generation or two to become popular. Especially if we are making lots of genetic manipulations in animals by then.
The problem with the Diamond Age Primer isn’t the communication and link with the child. Phones are that addictive already! The problem is the Primer is really smart, able to understand spoken language well and construct stories that appeal to the child. I think we’re nowhere near that yet.
And if you did have that level of AI, it would show up all over the culture. The Diamond Age doesn’t use AI anywhere else in the story that I recall.
Nick Bostrom, the machine intelligence/super human intelligence research professor, thinks embryo selection will be inevitable, because genome-wide association studies are inevitable (he is making no moral statement, only technological). 20-minute video, start at 5:30:
https://www.youtube.com/watch?v=-UIg00a_CD4
I agree with Mr. Pinker. I think genetically engineering babies will prove hugely complicated and impractical, beyond fixing the single-gene, point mutations.
Our genome is optimized for sexual attractiveness. And while fixing one particular trait (e.g., intelligence) may sound wonderful, it will likely come with side effects that render our young less attractive overall (e.g., autism). Scott Alexander has written interesting things about this topic. (http://slatestarcodex.com/2017/05/26/the-atomic-bomb-considered-as-hungarian-high-school-science-fair-project/)
You can’t change men without also changing women’s taste for men. Or put another way, a billion years of evolution will not easily be undone.