For what it’s worth I’m cautiously optimistic that “reverse-engineering the circuits underlying empathy/love/altruism/etc.” is a realistic thing to do in years not decades, and can mostly be done in our current state of knowledge (i.e. before we have AGI-capable learning algorithms to play with—basically I think of AGI capabilities as largely involving learning algorithm development and empathy/whatnot as largely involving supervisory signals such as reward functions). I can share more details if you’re interested.
For what it’s worth I’m cautiously optimistic that “reverse-engineering the circuits underlying empathy/love/altruism/etc.” is a realistic thing to do in years not decades, and can mostly be done in our current state of knowledge (i.e. before we have AGI-capable learning algorithms to play with—basically I think of AGI capabilities as largely involving learning algorithm development and empathy/whatnot as largely involving supervisory signals such as reward functions). I can share more details if you’re interested.