Goertzel on AGI existential risk, why P(doom) is a made-up number, what Anthropic and Dario Amodei are really doing, and why the slowdown will not happen. Summarized by section, with his actual words one click away.
Original post on XRecorded September 202610 sections · 234 transcript lines
The argument in five lines
P(doom) numbers are feelings dressed up as statistics. There is no data to forecast from, and nobody prices the risk of not building AGI.
Yudkowsky's inevitability argument fails because provable safety guarantees are impossible: you cannot formalize the world an AGI acts in.
Anthropic's alarm mixes genuine concern with IPO marketing and an elitist "we are the right ones to steer it" streak.
He agrees with Dario that scenario analysis beats probabilities, but not on the winning scenario. Dario wants a sanctioned lead; Goertzel wants open, decentralized development.
The slowdown will not happen. Neither Washington nor Beijing backs it, China and big US industry want open AI, and you cannot stop every smart person with a computer.