[ben_w]

On the one hand, based on FSD, Musk would be one of the first people to attempt this, well before the AI was capable enough to do a good job.

On the other, super-rich people can already afford a lot of humans to assist with raising their kids. AI isn't strictly necessary for this, as seen with e.g. Genghis Khan.

But yeah, if the G and the I in AGI is good enough and general enough, then robots directed by it could do a good job of raising kids — but at some point, you have to ask what life is even for, why you're having kids at all if you don't want to be involved in raising them.

[trhway]

> a lot of humans to assist with raising their kids. AI isn't strictly necessary for this, as seen with e.g. Genghis Khan.

it willn't be just kids. It will be "designer babies". Cleaned-up DNA with imports from Mozart and Einstein DNAs, and countless other improvements and enhancements for higher IQ, stronger muscles, faster reaction, better health, etc and cyber-enhanced with Neuralink style connection from may be even before being born, etc. There would be not many humans who would be able even to just keep up, less serve as teachers and trainers to such kids.

> to be involved in raising them.

engineering them is also involvement.

[ben_w]

> engineering them is also involvement.

Have you ever played video games with cheats enabled? Some of them are still fun, others feel pointless. If you want to treat your offspring as a machine to be engineered rather than raised, why bother with flesh in the first place, why not regard the AI itself as the offspring?

If a child were engineered as thoroughly as you suggest, the distance between them and their parents would necessarily be as severe as the distance between them and their potential human teachers.

But also there is a practical issue:

Musk specifically may well test such things on his own kids before the tech is actually ready, but any sensible person would want these interventions to have passed a full clinical trial before reaching their offspring. Such trials would necessarily lead to there being many humans with any one of those enhancements before all of them, making any gap much smaller in practice than you anticipate.

[trhway]

>to treat your offspring as a machine to be engineered rather than raised

raising is engineering, just very slow and with very limited results just because of the very limited toolset currently available. Look how parents using hormones and blockers, etc. - that is about max engineering available today and they do it because it benefits their children (i'm making pure engineering-wise point, and i'm not qualified nor have any intention to discuss whether it is really beneficial or not)

>why bother with flesh in the first place, why not regard the AI itself as the offspring?

The people replacing flesh children with AI would naturally be weeding their DNA out thus leading to the significant share of the population still preferring flesh ones.

>he distance between them and their parents would necessarily be as severe

parents want their kids to succeed, to do better than the parents. It is a biological imperative (as otherwise your DNA is weeded out by the one's who do get their children to do better, and thus those have been and will be the majority of the population). The first-gen immigrant parents working as say janitors and their college kids - huge difference and the parents are happy for it.

And don't forget that the parents here are themselves would already be the N-th generation of improvement, much ahead of the rest of the population.

>with any one of those enhancements before all of them

with any one, yet not all. Being a test subject for one feature is very different than having a package of those features.

>making any gap much smaller in practice than you anticipate.

in addition the test subjects may be prevented from propagating their DNA. (Cetaganda from the Vorkosigan saga :)