Funny how many things we learn about AIs also apply to humans, but sometimes they allow us to see the human thing from a new angle. Today I noticed “Corrigibility’s Desirability is Timing-Sensitive”, and the analogy is that you want human children to easily learn from the adults around them, especially from parents, but adult humans with the same level of openness would be too easy to brainwash; you actually want them to push back against the attempts to modify their values.
And that is why the rewiring of the pubescent brain involves changes that enable that. As the brain can’t hardwire changes to values, which are high-level learned, it has to go some other way.
Change the ground truth feedback (brain stem rewiring).
Weaken all previous connections (synaptic pruning?).
Reduce learning rate (myelination).
Something else? Changes to other hyperparameters (hormonal changes?).
Funny how many things we learn about AIs also apply to humans, but sometimes they allow us to see the human thing from a new angle. Today I noticed “Corrigibility’s Desirability is Timing-Sensitive”, and the analogy is that you want human children to easily learn from the adults around them, especially from parents, but adult humans with the same level of openness would be too easy to brainwash; you actually want them to push back against the attempts to modify their values.
And that is why the rewiring of the pubescent brain involves changes that enable that. As the brain can’t hardwire changes to values, which are high-level learned, it has to go some other way.
Change the ground truth feedback (brain stem rewiring).
Weaken all previous connections (synaptic pruning?).
Reduce learning rate (myelination).
Something else? Changes to other hyperparameters (hormonal changes?).