Update: DeepSeek confirmed working w PLA to avoid export controls.
https://x.com/andrewcurran_/status/1937141798893457470
Pretty sure the double -17 downvotes for trying to hold him accountable (to the question of whether or not he has updated) is proof of my original conjecture that this community has problems with epistemic humility.
Rationality is about logic, not tone or hurt feelings.
I thought the point of rationality was not to let our human emotions get in the way of truth.
I guess “racing” is bearing a lot of weight in your description.
Because (in part due to American worries about existential AI from folks on this message board), by some measures China is already “winning” the race.
So no it doesn’t have to be a dramatic centralized investment. There is a race, ongoing, and due to their commitment to open source, they may be winning.
again, it’s like our obsession with semantics are blinding us from the truth. Whether or not China has a massive centralized strategy, they have obviously caught up.
Whether or not you admit the behavior to your definition of “race” they have caught up.
Which would make me think you and Gwen would update, from prior views that there weren’t able to catch up (bc there was no evidence they were actively “racing”).
https://x.com/natfriedman/status/1928092126614597686
It’s a very simple question and speaks to the heart of the post, which he decided to comment on.
@Connor Leahy have you updated or not?
Saying you don’t like my tone is an ad hominem. It’s not rational.
ok cool, so no update!
Ladies and gentlemen, I rest my case!
FYI this was used yesterday, in this post.
https://www.aipanic.news/p/the-doomers-dilemma
maybe we can ask @gwern.
Gwern, at what point would you say you were 'wrong' and how would that make you 'update'?
what line would you agree on today @garrison ? At what point would you actually 'update'?
So far, am seeing a lot of people contesting the 'object' and not a lot of people updating. Which is kinda my point. Concordance with the group consensus seems to have become a higher priority than rationalism on this forum.
Before we get to the receipts, let's talk epistemics.
This community prides itself on "rationalism." Central to that framework is the commitment to evaluate how reality played out against our predictions.
"Bayesian updating," as we so often remind each other.
If we're serious about this project, we should expect our community to maintain rigorous epistemic standards—not just individually updating our beliefs to align with reality, but collectively fostering a culture where those making confident pronouncements about empirically verifiable outcomes are held accountable when those pronouncements fail.
With that in mind, I've compiled in the appendix a selection of predictions about China and AI from prominent community voices. The pattern is clear: a systematic underestimation of China's... (read 256 more words →)
”Baaaaah… at no point was China appreciably behind in generative AI. Not when ChatGPT was released and not now.
OpenAI prematurely released ChatGPT as a fund raising exercise. Three years and multiple iterations later, it’s still a fundraising exercise.
At the time, Tsinghua’s Wu Dao model was the most sophisticated (most parameters and multimodal well before its time). They just were not interested in releasing an uncommercial version into the market just to fund raise.
Wu Dao’s tech is now spread out in Huawei, Alibaba, DeepSeek, Kimi, Kling etc which are in now way behind US AI offerings… with investment spend a fraction of the US.
This Westoid AI circle jerking is just that. Westoids jerking each other off in one of the few tech fields they’re still competitive in… which is hoovering up resources from everything else.”
https://x.com/doggydog1208/status/1939247596201226601