-
safety
2025-09-19
In a book-release interview: any plan to fight something smarter than humanity is a dumb plan.
«you don't want a plan to get into a fight with something smarter than humanity. That's a dumb plan.»
-
regulation
2025-09-16
The book's policy demand: worldwide it must become illegal for AI firms to charge ahead with AI development.
«All over the Earth, it must become illegal for AI companies to charge ahead in developing AI.»
-
safety
2025-09-16
Alignment critique in the book: modern AIs are 'grown', not engineered, hence opaque and uncontrollable.
«The most fundamental fact about current AIs is that they are grown»
-
safety
2025-09-16
Central thesis of the 2025 book: whoever anywhere builds superintelligence kills everyone.
«If anyone anywhere builds superintelligence, everyone everywhere dies.»
-
safety
2025-09-12
Ahead of the book explains his alarmist stance: someone has to plainly say what is actually happening.
«Someone, at some point, just has to say what's actually happening.»
-
regulation
2025-03-17
Repeats the call to stop everything: we lack the technical ability to make superintelligent AI obedient and aligned.
«We must stop everything... We don't have the capability to design a superintelligent AI that is obedient and aligned.»
-
safety
2023-04-15 ~
At TED2023 compactly states: if we actually build superhuman AI with current methods, we all die.
«If we actually try to do this in real life, we are all going to die.»
-
safety
2023-03-29
Sharp extinction thesis: building a too-powerful AI now would kill every human and all biological life.
«every single member of the human species and all biological life on Earth dies shortly thereafter.»
-
regulation
2023-03-29
↺ muutti kantaa
In TIME demands not a pause but an indefinite global shutdown of AI development — a hardening from 'pause' to 'shut it down'.
«Shut it all down. We are not ready.»
