Scott Alexander
AI superforecasters have come close to the accuracy of top humans, with performance rising rapidly
There are plenty of good specific plans for how a global capabilities slowdown could be arranged
Open weights AI could be the difference between being free yeomen versus corporate serfs if AI becomes the linchpin of the future
In a year or two, either AIs will plateau at or slightly above human forecasting levels, or they will achieve ultraforecasting capabilities far beyond top humans