Why I reject the P(doom) framing, despite being very concerned about current AI trajectories

the P stands for Profit

Strongly agreed and feel the Anthropic report today is a great thing to remember when discussing all this. Tangible harms exist on a spectrum and just like infinity is incomprehensible to almost all to imagine because the concept can’t be held, so too is p(doom). Real harm hits harder and feels more tangible.

Exactly.

that guy was insufferable on the pod.

My extinction scenario is humans simply opting for voluntary oblivion... A few couples who can't have children digitise their DNA and have AI offspring. It catches on. Then biological children can't compete so everyone does it.

As a poker player you of all people know a single point estimate hides the whole distribution. This risk is overlapping failure modes with different timelines, each needing its own mitigations.

yes there is a 10% chance of an AI catastrophe, but if that catastrophe is avoided there is basically 100% chance of post scarcity society in 20 years. and that is before you get to humanoid robots. your garden shed could be a factory.

Humanity is worth saving

Seems reasonable, not sure what "everything we possibly can" means. Though, I feel the typical interpretation of this sounds good but leads to maximizing P(doom).

It's p(dystopia), not p(doom). The real concern is social manipulation, authoritarianism, spying, deceptions, cyber security, etc.

poker players do know about risk and can internalize it much better than general public and politicians and even programmers