Today I joined one of the video-conferences Alan calls his “salon”.

“All smart people should use IVF now,” Alan says, by way of introduction. “That’s what we are going to talk about today. Humans need to form a collective super-intelligence to manage the emergence of super-intelligent AIs.”

This speaker has a strong Scottish accent. He presents a table that shows the different intelligence gains resulting from different scenarios, that is, different levels of intervention on the creation of human embryos.

He starts with the example of the gains from selecting for intelligence among a sample of 10 human embryos: the potential gain is 11.5 IQ points. It’s a purely theoretical calculation, though, using a model derived from the variants in known samples of identical twins.

“The trouble is”, says the speaker, “that finding the genes for intelligence is proving a bit of a guddle.”

“What the hell is that?” asks Alan.

“Scottish slang for trying to catch fish with your bare hands…”

That gets a round of laughter.

”We already conduct PGD for over over 130 specific genetic defects. We don’t yet know all the genetic structures that promote intelligence but we are learning more all the time. Using the best current knowledge to select the best embryo from two very smart people will significantly increase the chances of getting a smart healthy child…”

PGD is pre-implantation genetic diagnosis. In IVF it is common to create ten embryos.

The speaker then talked about the problems of finding the genetic keys to intelligence. It’s turning out to be about a large network of genes in the right balance or combination. But minor variants in the same networks can be linked to inherited disease like Tay Sachs.

Alan said: “If we began to get reasonable pointers to intelligence they could be tested for. But everything else we are talking about will take 25 years to realise. That worries me…”.

A much more advanced technology, which could generate much faster intelligence gains but which is as yet purely theoretical, would derive egg and sperm cells from embryonic stem cells and fuse them to  produce a new embryo. This is called Iterative Embryo Selection (IES). It is already being used experimentally with animals and has produced fertile mice and what look like fertile human sperm cells.

Here the same table showed a potential gain of over 60 IQ points over five generations of IES and 130 over 10. Massive.

“How long would it take before we can do this?”, Alan asked.

“Maybe ten years”, said the scientist, “but there would be huge regulatory issues…”

Other options are discussed like pushing human brains in ways that our evolution may not have favoured. For example, artificially speeding up their metabolism to use more energy than our evolutionary environment could have afforded.

The speaker said IES would abolish the biological link between parents and their children. To avoid in-breeding and optimise the result, each embryonic stage could be enhanced from new sources. So a parent would be a “social” rather than a “biological” parent.

I thought about Mikey and Jamie and Hadley. How important is it to know a child has some of your own genes? Would it be enough to know that somewhere in the process your own genes had played their part?

Would it be different for the woman? She bears the child.  

After the scientist had been dismissed with thanks, Alan said: “What are the ethics here? Maybe, in secret, some regimes are already trying to produce super-intelligent kids…”