A personal guess about the chance that AI could cause a human disaster.
It is like rating how risky a road trip feels. Saying 10% means, “We will probably get there. But let's check the brakes first.”
People use it to debate AI safety rules and spending. You may hear it in AI labs and government talks.
Alignment
P(doom) estimates disaster risk from poor Alignment.
AGI
People use P(doom) to discuss risks after AGI arrives.
Superintelligence
People often give a higher P(doom) for Superintelligence.