Can you elaborate on that? It seems non-obvious; I feel like I could tell a different story
… like, OpenAI has much fewer resources than Deepmind, so if Deepmind released GPT-3 first, and if it is the case that scaling up a language model far enough gives you AGI, there would have been less time in between knowing what AGI will look like and having AGI.
Releasing GPT-3 non-trivially increased the odds of doomsday. So yeah they are not good actors.
Can you elaborate on that? It seems non-obvious; I feel like I could tell a different story
… like, OpenAI has much fewer resources than Deepmind, so if Deepmind released GPT-3 first, and if it is the case that scaling up a language model far enough gives you AGI, there would have been less time in between knowing what AGI will look like and having AGI.