If TAI arrives and doesn’t cause extinction, it could still be years before the poorest countries are significantly impacted. So, the probability of TAI arrival could be too high to discount by (or at least for AI’s contribution).
Also, a life saved might become more valuable and life-saving charities might do more good than otherwise, in case the beneficiaries’ quality of life or life expectancies improve due to the arrival of TAI! I’d guess you’d only want to discount by the probability of extinction or global catastrophe for life-saving interventions. I suppose there’s also a chance that between your donation and its use saving a life, the beneficiary would have been saved through the benefits from TAI, but I think GiveWell has been recommending donations for benefits in the next couple years after reception, so this seems unlikely.
The extreme person-affecting tails involve far longer lives from life extension tech and mind uploading.
Income/wealth gains would probably become less valuable if GiveWell charity beneficiaries benefit from TAI.
All good points. Yes, in slower take-off scenarios there would be a larger lag, I suppose I was implicitly thinking of cases where the world quickly moves to collapse or >=20% annual economic growth, but true this does weaken my conclusion. Ah interesting thought about saving lives being especially valuable given the possibility of life-extension tech. Perhaps our best guess ‘life expectancy’ for someone alive today should be >100 years then and maybe far more, if there is even a small chance of entering post-death worlds.
I think it requires either a disagreement in definitions, or very pessimistic views about how tractable certain scientific problems will prove to be, to think that the “transformative” bit will take long enough to impact the discount rate by more than a few percent (total). But yes, it will be non-zero.
If TAI arrives and doesn’t cause extinction, it could still be years before the poorest countries are significantly impacted. So, the probability of TAI arrival could be too high to discount by (or at least for AI’s contribution).
Also, a life saved might become more valuable and life-saving charities might do more good than otherwise, in case the beneficiaries’ quality of life or life expectancies improve due to the arrival of TAI! I’d guess you’d only want to discount by the probability of extinction or global catastrophe for life-saving interventions. I suppose there’s also a chance that between your donation and its use saving a life, the beneficiary would have been saved through the benefits from TAI, but I think GiveWell has been recommending donations for benefits in the next couple years after reception, so this seems unlikely.
The extreme person-affecting tails involve far longer lives from life extension tech and mind uploading.
Income/wealth gains would probably become less valuable if GiveWell charity beneficiaries benefit from TAI.
All good points. Yes, in slower take-off scenarios there would be a larger lag, I suppose I was implicitly thinking of cases where the world quickly moves to collapse or >=20% annual economic growth, but true this does weaken my conclusion. Ah interesting thought about saving lives being especially valuable given the possibility of life-extension tech. Perhaps our best guess ‘life expectancy’ for someone alive today should be >100 years then and maybe far more, if there is even a small chance of entering post-death worlds.
I think it requires either a disagreement in definitions, or very pessimistic views about how tractable certain scientific problems will prove to be, to think that the “transformative” bit will take long enough to impact the discount rate by more than a few percent (total). But yes, it will be non-zero.