A pause is a resource though. You’re giving them state enforcement of their preferred development schedule. That much is an objective statement (or it cruxes on how much Anthropic supports that rendition of the pause.)
I don’t think the labs agree with the message. Jacob quit. Hubinger did not. Whether or not this is worth quitting over seems like a serious crux, not a slight difference in opinion. Jacob probably favors slowdown legislation that OpenAI and Anthropic won’t agree with. You’ll know because they’ll switch to their second pacing prong, “it lets China win.”
I can be more precise, “they are trying to capitalize on the message.” I’d like them to fail. Ideally the public consciousness reaction would be, “wait, Jacob and Anthropic must want different things, despite superficial overlap in their message.”
Just to clear my overall stance, I think OpenAI and Anthropic are authoritarian-takeover threats. I am paying attention to how much power they’re amassing not just from their tech but from the U.S. government. For example at some point they’ll argue Chatgpt Surveillance (tm) is the only way to stop bio terrorists (they’ll fail to stop their own AI doing it and will have poor visibility into China, of course.)
In fact I think x-risk skeptics are forming some firewall against s-risk. Maybe the modal rationalist stance is to not distinguish too much, because slowing the most powerful labs and slowing the tech are the same right now. Maybe we would de-couple some once serious “pacing” legislation is produced.
A pause is a resource though. You’re giving them state enforcement of their preferred development schedule. That much is an objective statement (or it cruxes on how much Anthropic supports that rendition of the pause.)
I don’t think the labs agree with the message. Jacob quit. Hubinger did not. Whether or not this is worth quitting over seems like a serious crux, not a slight difference in opinion. Jacob probably favors slowdown legislation that OpenAI and Anthropic won’t agree with. You’ll know because they’ll switch to their second pacing prong, “it lets China win.”
I can be more precise, “they are trying to capitalize on the message.” I’d like them to fail. Ideally the public consciousness reaction would be, “wait, Jacob and Anthropic must want different things, despite superficial overlap in their message.”
Just to clear my overall stance, I think OpenAI and Anthropic are authoritarian-takeover threats. I am paying attention to how much power they’re amassing not just from their tech but from the U.S. government. For example at some point they’ll argue Chatgpt Surveillance (tm) is the only way to stop bio terrorists (they’ll fail to stop their own AI doing it and will have poor visibility into China, of course.)
In fact I think x-risk skeptics are forming some firewall against s-risk. Maybe the modal rationalist stance is to not distinguish too much, because slowing the most powerful labs and slowing the tech are the same right now. Maybe we would de-couple some once serious “pacing” legislation is produced.