Alibaba has become the first Chinese company to make Meta’s Llama 2 available as it started hosting the LLM on its cloud computing platform earlier this week.
Making the announcement on its WeChat account, Alibaba said, “Today, Alibaba Cloud has launched the first training and deployment solution for the entire Llama 2 series in China, welcoming all developers to create customized large models on Alibaba Cloud.”
When Meta announced that Llama 2 would be released for free, it explained that it hoped being free and “open-source” would lead to greater adoption and testing of the model.
That’s probably a smart business decision when you’re trying to get into a largely untapped Chinese market of more than a billion people. While Alibaba will almost certainly need to pay Meta to use its LLM, it’ll likely be a lot less than the pricey proprietary models that power ChatGPT.
With Alibaba coming under increased scrutiny from the Chinese authorities, the deployment of Meta’s Llama 2 raises some sticky issues.
Well this is awkward
Meta is probably right to be optimistic that Chinese companies will be eager to adopt its technology in their products. The irony is that other Meta products like Facebook, WhatsApp, and Instagram have been banned in China for years now.
China’s unabashed control over society and how it communicates online is well known, so the deployment on Alibaba’s servers would not have gone unnoticed. The debatable open-source nature of Llama 2 is no doubt also appealing to Chinese AI regulators who would certainly want to peer under the hood.
Meta has been criticized for making their powerful AI available to the wild in the manner that they have, and it brings an interesting dimension to the East vs. West AI race. It also raises some ethical issues as users of Llama 2 on Alibaba will have to ensure that the model’s output “aligns with the core values of socialism” as defined by the Chinese state.
The US has tried to rein in the speed at which China is making advances in AI by increasingly looking to place export controls on AI semiconductors. Despite that, Chinese companies like Alibaba, Huawei, Tencent, and Baidu have made advancements with the release of their own generative AI models.
That Alibaba has embraced Llama 2 instead of sticking solely with its own Tongyi LLMs may be indicative of the comparable advantages that Llama 2 brings to the game.
Having cutting-edge American AI technology deployed in China will certainly have some tongues wagging in the US Congress. It’s also interesting to see the different approaches the leading AI companies have been taking.
Anthropic, Google, Microsoft, and OpenAI have just launched the Frontier Model Forum. It’s an industry body focused on ensuring the safe and responsible development of advanced AI models. Meta was noticeably absent from the list of founding companies.
