Alibaba has announced a new update for its AI model Tongyi Qianwen, bringing in hundreds of billions of parameters, poising it to become one of the world's most powerful AI models.
According to Reuters, eight industry-specific AI models were also announced during the recent Chinese tech giant's annual conference in Hangzhou.
Alibaba's cloud computing arm noted that its Tongyi Qianwen 2.0 has "hundreds of billions of" parameters, which is a benchmark used to measure the power of an AI model.
For comparison, Reuters reported that back in 2020, the ChatGPT-3 AI model from Microsoft-backed OpenAI had 175 billion parameters, while Meta Platform's Llama 2 model had 70 billion parameters this year.
The eight AI models announced by Alibaba are reportedly for the entertainment, finance, healthcare, and legal industries, with dedicated tools for generating graphics, writing computer code, analyzing financial data, and legal paper searches. These tools went online on Tuesday following the announcement.
(Photo : Andrea Verdelli/Getty Images)
A group of experts has recognized the need to regulate and govern artificial intelligence (AI) effectively.
Alibaba AI Model Tongyi Qianwen
Forbes described Alibaba's Tongyi Qianwen as a question-and-answer chatbot that is also capable of being a coding co-pilot, generating and automatically error-checking code in several computer programming languages, much like ChatGPT and its other competitors.
Tongyi Qianwen is also one of the few AI models that are open-sourced or source code freely available to the public and may be updated by anybody.
This has yet to be the case with other models, such as OpenAI's GPT-4. Open-sourcing technology has various advantages for its inventors since it allows it to integrate enhancements, changes, or re-engineering performed by developers worldwide.
Reuters previously reported that Alibaba Cloud Intelligence Division partnered with organizations such as OPPO, Taobao, DingTalk, and Zhejiang University to train their own large language models or develop language model applications based on Tongyi Qianwen.
During the announcement, Alibaba chairman Joe Tsai said about half of large-language AI models in China currently run on Alibaba Cloud. According to Tsai, Alibaba's AI community ModelScope now includes 2,300 models and 2.7 million developers.
Alibaba CEO Eddie Wu has previously said that his company's two top focuses would be to become more customer-centric and AI-driven.
Read Also: China May Blacklist Illegal AI Training Data
China's AI 'War of a Hundred Models'
The move comes amid China's current "War of a Hundred Models," a term created by Tencent vice president Jiang Jie, which describes the current rush to design and deploy generative AI in the Chinese sector.
The AI "war" in China has seen a frenzy of product announcements from startups and IT behemoths almost daily, reportedly triggered by ChatGPT's success. It also prompted Chinese tech giants, like Alibaba and Huawei, to promote their offerings.
Alibaba's rival, Tencent, has also joined the race, saying its Hunyuan AI model has more than 100 billion parameters and surpasses OpenAI's GPT-4 model in Chinese processing.
Data from brokerage CLSA reportedly showed that China now has at least 130 large language models, comprising 40% of the global total and behind the United States 50%.
Related Article: US-China Tech War: China's Zhipu AI Raises $340 Million to Fuel AI Advancements Amid Chip Restrictions
