KOMPAS.com – Chinese firm Alibaba on Tuesday, April 11, introduced Tongyi Qianwen, an AI-powered language model similar to GPT. At the same time, Beijing revealed a draft law that new AI must "reflect socialist values."
The new AI model Tongyi Qianwen — which means "truth from a thousand questions" — can draft invitation letters, summarize meeting notes and advise users on what type of makeup to purchase, as shown in a filmed demonstration on Tuesday.
"We are at a technological watershed moment driven by generative AI and cloud computing, and businesses across all sectors have started to embrace intelligence transformation to stay ahead of the game," said Alibaba Group CEO Daniel Zhang.
The model will be integrated into DingTalk, Alibaba's workplace messaging app, and Tmall Genie, its voice assistant. Tongyi Qianwen can draft business proposals and emails and "will bring about big changes to the way we produce, the way we work, and the way we live our lives," according to Zhang.
Last Friday, Alibaba published a teaser for the new chatbot, with a post on social media reading: "Hello, my name is Tongyi Qianwen, this is our first time meeting, I welcome your feedback."
Alibaba Cloud plans to open Tongyi Qianwen to customization from users so that they can build their own large language models.
The AI chatbot race
Generative AI is a type of artificial intelligence that can create original content, such as text or images, by learning from the data it is trained on.
Interest in generative AI has spiked since the Microsoft-backed OpenAI launched the chatbot sensation ChatGPT last year.
Since then, rival companies have rushed to launch similar products, such as Google's Bard and Baidu Inc's Ernie Bot. On Monday, Chinese AI company SenseTime also revealed a chatbot called SenseChat along with a range of new AI products.
Stricter regulation on AI in China
Also on Tuesday, China's Cyberspace Administration (CAC) revealed draft measures for regulating generative AI services, as more companies like Alibaba introduce publicly available AI models.
The regulator wants companies to submit AI products for a "security assessment" before being released to the public.