For most people in their 20s or 30s, it is quite unlikely (around 10%) that they die before AGI. And if you basically place any value on the lives of people other than yourself, then the positive externalities of working on AI safety probably strongly outweigh anything else you could be doing.
Acceleration probably only makes sense for people who are (1) extremely selfish (value their life more than everything else combined) and (2) likely to die before AGI unless it’s accelerated.
Most longevity researchers will still be super-skeptical if you say AGI is going to solve LEV in our lifetimes (one could say—a la Structure of Scientific Revolutions logic—that most of them have a blindspot for recent AGI progress—but AGI=>LEV is still handwavy logic)
Last year’s developments were fast enough for me to be somewhat more relaxed on this issue… (however, there is still slowing core aging rate/neuroplasticity loss down, which acts on shorter timelines, and still important if you want to do your best work)
Another thing to bear in mind is optimal trajectory to human immortality vs expected profit maximizing path for AI corps At some point, likely very soon, we’ll have powerful enough AI to solve ageing, which then makes further acceleration very -ve utility for humans
I don’t know whether to believe, but it’s a reasonable take...
For most people in their 20s or 30s, it is quite unlikely (around 10%) that they die before AGI. And if you basically place any value on the lives of people other than yourself, then the positive externalities of working on AI safety probably strongly outweigh anything else you could be doing.
Acceleration probably only makes sense for people who are (1) extremely selfish (value their life more than everything else combined) and (2) likely to die before AGI unless it’s accelerated.
“10% is overconfident”, given huge uncertainty over AGI takeoff (especially the geopolitical landscape of it), and especially given the probability that AGI development may be somehow slowed (https://twitter.com/jachaseyoung/status/1723325057056010680 )
Most longevity researchers will still be super-skeptical if you say AGI is going to solve LEV in our lifetimes (one could say—a la Structure of Scientific Revolutions logic—that most of them have a blindspot for recent AGI progress—but AGI=>LEV is still handwavy logic)
Last year’s developments were fast enough for me to be somewhat more relaxed on this issue… (however, there is still slowing core aging rate/neuroplasticity loss down, which acts on shorter timelines, and still important if you want to do your best work)
https://twitter.com/search?q=from%3A%40RokoMijic%20immortality&src=typed_query
I don’t know whether to believe, but it’s a reasonable take...