Note these forecasts were all conditional on no transformative AI arriving before their resolution.
This is an important caveat. I think it’s a defensible/correct choice for us to focus on that for our forecasts[1]. However, it was a communication mistake of mine to not mention it more prominently earlier on, rather than buried in the “Methods” section.
EDIT: Thanks to Neil for fixing this error despite being on vacation! This is now displayed prominently in “Key Takeaways”
I take responsibility for the communications mistake. Because my work is more in the forecasting/longtermist/x-risk side of things, I should’ve realized that this post will be misleading to many people who might have thought these forecasts are an all-things-considered probability.
Thanks to simeon_c for pointing out this mistake (also mentioned on Twitter).
[1] Roughly speaking, the primary target audience for who those forecasts are meant to inform are funders looking into investing in cultured meat, and a secondary target audience are individuals looking into working in CM. Since worlds where AGI will make this problem irrelevant are not worlds where you want to invest in CM (even/especially if AGI can make it happen much more quickly).
This is an important caveat. I think it’s a defensible/correct choice for us to focus on that for our forecasts[1]. However, it was a communication mistake of mine to not mention it more prominently earlier on, rather than buried in the “Methods” section.
EDIT: Thanks to Neil for fixing this error despite being on vacation! This is now displayed prominently in “Key Takeaways”
I take responsibility for the communications mistake. Because my work is more in the forecasting/longtermist/x-risk side of things, I should’ve realized that this post will be misleading to many people who might have thought these forecasts are an all-things-considered probability.
Thanks to simeon_c for pointing out this mistake (also mentioned on Twitter).
[1] Roughly speaking, the primary target audience for who those forecasts are meant to inform are funders looking into investing in cultured meat, and a secondary target audience are individuals looking into working in CM. Since worlds where AGI will make this problem irrelevant are not worlds where you want to invest in CM (even/especially if AGI can make it happen much more quickly).