"A few thousand days"
Sam Altman places superintelligence around 2032–2035.
Intelligence beyond the best humans at everything — including building the next version of itself. Past this point forecasting is pointless: the pace is no longer set by people.
Extrapolation stops working here. No single branch leads to this point alone: continual learning, autonomous science and verifiability are all required at once.
Sam Altman places superintelligence around 2032–2035.
Dario Amodei argues human-level is within a few years; Demis Hassabis puts genuine AGI a few years out, not here.
The mechanism this point is about: a system accelerating its own development. It turns progress from linear into unpredictable.
The higher the capability, the more a goal mismatch costs. Not philosophy but an engineering requirement on the reliability gate.