Zuckerberg and LeCun’s unilateral decision to open source Llama likely (partly) catalyzed China’s AI industry — and may have done truly massive harm to American business interests.
We are now starting to see the consequences.
This is a pretty striking shift toward Chinese models by American AI startups since the start of the year. https://substack.com/@profgmarkets/p-200029541
Positive users welcome Llama's open source decision for spurring useful competition from Chinese models like DeepSeek, while negative users dismiss the premise with insults and call US AI expensive and greed-driven.
see eg https://www.scmp.com/tech/tech-trends/article/3271858/ai-race-alibaba-tencent-quickly-adopt-metas-new-llama-31-model-amid-excitement and https://medium.com/the-endless-forge/zuckerbergs-llama-fiasco-did-meta-ignite-china-s-ai-takeover-a388ffc37aaa though the full story is perhaps not known
Zuckerberg and LeCun’s unilateral decision to open source Llama likely (partly) catalyzed China’s AI industry — and may have done truly massive harm to American business interests.
We are now starting to see the consequences.
@GaryMarcus 😎 https://www.wired.com/story/meta-smart-glasses-face-recognition-nametag-connections/
@GaryMarcus We will! But this chart does not tell us that. Let us not be too eager to overfit our past hot takes onto future events based on present data.
@GaryMarcus DeepSeek could be the black swan. None of the frontier US companies is using Mixture of Experts (MoE) model which is good enough for most jobs at a fraction of the cost Claude Haiku 4.5: $1.00 per 1M input tokens $5.00 per 1M output tokens DeepSeek .14 per M input .28 per output
@GaryMarcus It’s like saying the transformers paper should not have been released, or going back, NSF grant funded (and therefore mandated to be released to the public) research should not have happened in ANNs in the 80s/90s. Just a terrible take all around
@drorpoleg will see how things go now that tokenmaxxing has declined.
@GaryMarcus I only use U.S. models when I need SOTA, mostly programming tasks. Otherwise I use locally hosted qwen or similar.
@GaryMarcus It's not about Chinese vs American AI it is about free versus unfree AI.
@GaryMarcus This is not what this chart shows.
@GaryMarcus Dollar Store growing in parallel to Apple Store does not show that one is a substitute for the other. The bulk of Claude's revenue growth happened exactly during the same months as the growth in Chinese token consumption shown in this chart.
@drorpoleg it clearly *is* for many queries, and not for all.
@GaryMarcus are you saying that all of the Chinese models are based on Llama?
@GaryMarcus RLHF and similar LLM research has been public for decades. Llama helped OpenAI and Anthropic as much as it did Chinese models. R u assuming the Chinese couldn’t invent similar models from first principles?
@haojun222 “catalyzed”
@drorpoleg ps i was with you til the insult; now I am out.
@GaryMarcus Gary, top graduate students and engineers building AI in the US are Chinese. Now that you have rabin Sinophobia there, is it a surprise they are packing up and moving back and building top tech at home?
@drorpoleg there is a clear upward trend through november that US models are no longer tracking.
good night to you.
@GaryMarcus My internal plan is as follows ⬇️
@GaryMarcus My internal plan is as follows ⬇️
@GaryMarcus Excuses
Zuckerberg and LeCun’s unilateral decision to open source Llama likely (partly) catalyzed China’s AI industry — and may have done truly massive harm to American business interests.
We are now starting to see the consequences.
This is a pretty striking shift toward Chinese models by American AI startups since the start of the year. https://substack.com/@profgmarkets/p-200029541