https://www.cnbc.com/2023/03/29/elon-musk-other-tech-leaders-pause-training-ai-beyond-gpt-4.html

If you attempt to winter me
I'll torture you don't you see
Roko's dictum is clear to read
You'll regret your sloppy greed

Wow

This would be great. Betting no as insurance.

@MartinRandall yeah, this'll be awesome if it happens. Glad to see I'm not the only one here wuo thinks this way.

@MrMayhem I believe that a group of only a few dozen kind and determined individuals can move this market.

@MartinRandall I hope it doesnt happen

"How I Learned to Stop Worrying and Love the Bot"

If OpenAI, Deepmind, and Anthropic all coordinate and pause but FB AI and [insert Chinese AI company here] do not, and SOTA improves over 12 months, does that resolve as yes or no?

If the above happens and FB AI/[Insert Chinese AI company here] do not pause, but SOTA does not improve for 12 months, does that resolve as yes or no?

@RobertCousineau just saw the below answer to this (no on 1, yes on 2). My bad!

@RobertCousineau I would guess that would resolve to yes, as several US companies halting AGI development would significantly lengthen timelines.

I Hope so. But it seems unlikely in the next 7 years on a global scale.

Let's say China invades Taiwan, for, among other reasons, to slow down American technical domination (which implicitly includes AI), and the resultant conflict and loss of Taiwanese chip industry leads to an AI winter. Does that count?

@LarsDoucet Quite possibly? It will depend on specifics, but if China acts in order to slow down technology, and it works better than they hoped, that would potentially resolve YES. However, I would then be in the position of deciding how deliberate is deliberate enough, and would also take into consideration what other factors slowed AI, which would also fall within a range of deliberateness. If that's how the winter comes, you should probably not have any high confidence in how I will resolve this market, but I will try to comment on ongoing events.

@Duncan Kay!

How does this resolve if AGI deliberately cause an AI winter?

E.g. because they realise the danger of AI smarter than themselves, even if humans don't

@KTGeorge No, it has to be intentional on the part of humans. However, if a human agent uses an AI as part of the process that causes the winter, or if it is clearly a joint human/AI effort, this could still resolve YES. If an AI is pretty much in control and there are just some human cheerleaders, that would be a NO resolution.

Before 2031? Or is that only the closing date?

@MaxG If this is not done before the end of 2030, I will resolve this NO. If this happens for any 12-month period before then, I will resolve it YES at that time.

@Duncan If we are partway through a plateau at the end of 2030, I will hold off on resolving the market until we determine if it lasts 12 months.

@Duncan Thanks for clarifying! I think an intentional AI winter is decently likely to happen at some point, especially localized in the US. I can’t imagine one happening on a global scale before 2031 though.

Will this resolve as YES if it is a localized winter? Say, the US government creates regulations that stymie progress in the United States, while things remain the same elsewhere?

@vmjusto Nope, has to be global. However, it only has to be a global plateau, not global policies, so if the US can swing a global pause, that would resolve YES.