Does continued AI development make superintelligence—and certain doom—inevitable?
A poll asks builders of large language models listed on Artificial Analysis to weigh in, voting according to the side of the training pipeline they mainly work on.
TLDR
A poll asks whether continued AI development inevitably leads to a “Yudkowsky-style superintelligence” with P(doom)=1—a 100% probability of doom. It invites people who build large language models appearing on Artificial Analysis to vote based on the side of the training pipeline they mainly work on.
Does continued AI development make superintelligence—and certain doom—inevitable?
A poll asks builders of large language models listed on Artificial Analysis to weigh in, voting according to the side of the training pipeline they mainly work on.
TLDR
A poll asks whether continued AI development inevitably leads to a “Yudkowsky-style superintelligence” with P(doom)=1—a 100% probability of doom. It invites people who build large language models appearing on Artificial Analysis to vote based on the side of the training pipeline they mainly work on.