In response to the foreign development of powerful chatbots like ChatGPT, the Chinese giant Alibaba late last week unveiled its generative AI model dubbed Tongyi Qianwen (which translates to “seek the truth through a thousand questions”).
According to the company, Tongyi Qianwen will initially be incorporated into Tmall Genie smart speakers and local messaging app DingTalk. In a second moment, AI should be applied to all Alibaba services, including e-commerce — a segment for which it is best known by Brazilians because of Aliexpress.
The “Chinese ChatGPT”
According to the company’s presentation, Tongyi Qianwen was trained on a vast amount of data to produce compelling responses to user requests in both English and Mandarin. Among the tasks that Alibaba’s AI can perform are: summarizing meeting notes, composing emails, composing business proposals, telling stories, planning trips and giving shopping advice.
During the presentation of the new artificial intelligence language in Beijing, the CEO of the Alibaba group, Daniel Zhang, said that “we live in a watershed moment in the technology field, driven by generative AI and cloud computing.” The company also has a major business division in cloud computing and envisions the use of AI in that sector as well.
AI and China’s regulatory policy
In recent months, Chinese giants are creating AI-based solutions to compete with foreign platforms. In March, Baidu introduced its 10 billion parameter language model, ERNIE-ViLG. The IDEA lab, run by former Microsoft Harry Shum, has unveiled Taiyi, a large-scale text-to-image model. SenseTime, a prominent AI company using computer vision, has launched the SenseChat chatbot.
Despite progress in the AI sector, the news comes at a troubled time for Chinese business. Beijing’s strict policy to regulate the sector, with hefty fines for entrepreneurs who do not respect competition rules, has hurt some areas of technology. In this scenario, some businessmen are resigning or even seeking refuge abroad.
In addition, China faces other challenges, of a geopolitical nature, which further complicate the sector. This is the case of the US embargo on cutting-edge technologies, which has a special impact on the importation of semiconductors — essential for the development of new technologies. China is trying to get around this situation internally with subsidies and price reductions for domestic components, but the solution has also been the target of criticism.
This conjuncture of factors explains the comment of the Minister of Science and Technology of China, Wang Zhigang, about the delay of Chinese AIs in relation to foreign models: “the important thing is to dribble and shoot, but it is not easy to be as good as Messi.” According to Wang Zhigang, “ChatGPT is very complex in terms of technological progress, especially in ensuring the real-time performance and efficiency of the algorithm. OpenAI has its advantages in that regard.”
Added to this is the regulatory tightening of advances in artificial intelligence. The Cyberspace Administration of China announced in early April draft rules to inspect the safety of generative AIs prior to their application and commercialization of products.
According to the rule, which is not yet scheduled to come into force, content generated by AI “must reflect fundamental socialist values and must not contain content related to the subversion of state power”. The goal, according to the regulatory agency, is to ensure the healthy development and standardized application of generative AIs.
