akira (林央, Hayashi Akira)’s p(doom)
In their words
akira (林央, Hayashi Akira) (Chief Alignment Officer / principal researcher, AGI Hub (Japan); University of Tokyo undergraduate) put their p(doom) at ~90% overall (all-things-considered probability of human death/extinction from AI ("90%ぐらいの死亡"); the 99.9% figure he usually quotes is conditional on no development regulation and superintelligence arriving within 5 years, not stated (conditional figure: AGI within 5 years)), on video on Sep 18, 2026.
Let me just say one thing first: when I say something like 99.9%, I'm actually talking under the condition 'no regulation happens and it gets built within five years'. If you factor in development regulation happening, or superintelligence actually arriving later, then overall it's about 90% death.
Original: ちなみに先に一つだけ言わせていただくと、僕が99.9%みたいに言う時は「規制ができなくて5年以内に出来ちゃった場合」という条件で実は話していて、開発規制ができたりとか、実はもっと超知能ができるのが先みたいなのを考慮すると、全体としては90%ぐらいの死亡っていう感じですね。
Source · re-checked against the primary source.
AGI Hub. His usual "99.9%" is conditional on no regulation and superintelligence within 5 years; overall ~90%. Date is the transcript's publication date; the livestream date is unconfirmed.
In context
Among safety orgs in the database (28 people), the median is 50%. Across all 223 people it is 10%. How the medians and the index are built.
A p(doom) is a personal, subjective estimate of the probability that advanced AI ends in human extinction or a comparable, irreversible catastrophe. Scope and timeframe differ from person to person, so compare with care: why p(doom) numbers disagree.