As to superhuman AI, I think Tamay Besiroglu and the people at OpenPhil (now Coefficient Giving) had papers that explored the possibility of explosive growth under circumstances like this. The arguments seemed counterintuitive but compelling.
I think some of those percentile ranges should just be considered mistaken? I agree that there are many considerations at play but, for example, the labour force participation rate question under (their incredibly strong definition of) rapid AI progress doesn't make sense.
I shared more information in a restack note, where I could format things properly. Would be interested to know what I'm missing.
On Imas’ point we might already seeing novel firms of the sort he’s talking about: https://www.nytimes.com/2026/04/02/technology/ai-billion-dollar-company-medvi.html?unlocked_article_code=1.X1A.TZcY.c7G_whUqLFQY&smid=url-share
As to superhuman AI, I think Tamay Besiroglu and the people at OpenPhil (now Coefficient Giving) had papers that explored the possibility of explosive growth under circumstances like this. The arguments seemed counterintuitive but compelling.
I think some of those percentile ranges should just be considered mistaken? I agree that there are many considerations at play but, for example, the labour force participation rate question under (their incredibly strong definition of) rapid AI progress doesn't make sense.
I shared more information in a restack note, where I could format things properly. Would be interested to know what I'm missing.