AI milestone: China authorises initial batch of LLMs in 2023
China has given its clearance to the first group of LLMs for this year, following the legalisation of generative AI services in August last year.
The authorizations arrive as the United States considers restricting Chinese access to American cloud service providers, in order to prevent the country from developing sophisticated LLMs.
China has authorised more than 40 government-sanctioned LLMs and related applications.
Following ChatGPT’s impressive achievement, China made the decision to move forward in the field of AI by creating a government organisation focused on this technology. Given the responsibility of establishing a nationwide guideline for large language models (LLMs), which are important elements in training AI chatbots like ChatGPT, this effort highlights Beijing’s strategic approach to reduce possible interruptions while utilising the revolutionary capabilities of AI in traditional sectors. In July 2023, a recently established China Electronic Standardisation Institute, which operates under the Ministry of Industry and Information Technology (MIIT), started working on creating a domestic standard for LLMs.
Revealed at the World Artificial Intelligence Conference (WAIC) in Shanghai, the Institute recruited major technology companies Baidu, Huawei, 360 Security Technology, and Alibaba to lead a task team in developing this new LLM standard. The drive for standardisation reflects the acknowledgment by local authorities of the potential of AI to stimulate economic growth and function as a useful everyday tool.
At the same time, it underscores their careful attitude towards the related dangers, stressing the importance of controlled utilisation of this revolutionary technology.
The first-ever decision on generative AI
In April 2023, Beijing proposed a preliminary rule for generative AI services, which later became a legal requirement on August 15. The ultimate rule is more advantageous than the initial suggestion, highlighting growth and safety. It promotes creativity in generative AI and takes a tolerant and cautious attitude to regulation. Under the new regulations, all AI content services that generate text, photos, audio, and video for the Chinese public will be required to comply with these rules.
The revised rules demonstrate a friendlier attitude, removing harsh language from the previous version and indicating a favourable position towards creative progress in generative AI. Shortly after the law came into force, China entered a new period by granting clearance to its first set of generative AI services on August 31, 2023.
Prominent technological companies, such as Baidu and SenseTime, have been authorised to release chatbots similar to ChatGPT for public use. Baidu, SenseTime, Zhipu AI (a company financed by the state), and Baichuan, a new venture supported by Tencent Holdings and helmed by Sogou co-founder Wang Xiaochuan, then together announced the removal of invite-only or beta testing labels from their AI chatbots in separate statements.
Importantly, their creative generative AI services, demonstrated by SenseChat from SenseTime situated in Shanghai, have been easily available on their websites or can be downloaded through Chinese app stores.
China aims to promote AI
After months of strict oversight, China is now promoting the adoption of AI across its enterprises and population. During the previous six months, the country has given approval for 40 LLMs and AI applications, with a significant group of 14 receiving approval this week.
Simply put, Beijing is encouraging the citizens of China and its small businesses to adopt and utilise AI on a large scale. The previous strict control over available LLMs and AI applications has relaxed, resulting in a wide variety of LLMs entering the public domain. Actually, China’s recent authorization of the initial group of LLMs this year represents a notable advancement in the country’s AI field.
Unlike prior approvals that included GenAI models, this latest group includes various industry-specific LLMs. This illustrates the focused efforts in mainland China to promote AI development and regulation, with the goal of improving organisational efficiency.
According to reports from local media and company announcements, a total of 14 LLMs and enterprise applications have recently been approved for commercial use by authorities. The approved applications include those from smartphone company Xiaomi, AI specialist Beijing Fourth Paradigm Technology Co, and tech unicorn 01.AI, which was founded by venture capitalist Lee Kai-fu, states a report from South China Morning Post.
SCMP also mentioned some of the recent authorizations, such as startup Frontis.AI for its industry-specific LLMs – Pinshang and Moxiaoxian. These LLMs offer consumer goods companies with knowledge and assistance for product and marketing innovation. Frontis.AI intends to provide more customised services to its corporate clients. In addition, the XGPT AI model from cybersecurity firm ThreatBook, which is the first LLM focused on internet security in China, and the LLM from video solutions provider XinYi Tech, which specialises in AI-powered video production and processing tools, were authorised this month.
According to SCMP, ThreatBook’s LLM is designed to assist businesses in examining possible security risks, while XinYi Tech’s LLM focuses on advertising, education, media, and e-commerce apps. Notable AI applications authorised in January include a smart resume-enhancing tool from Chinese online job search portal Zhaopin, chatbots from e-commerce services company Beijing Zhidemai Tech, and a mobile provider of books, magazines, and comics called iReader Technology.
Stay Updated about the latest technological developments and reviews by following TechTalk, and connect with us on Twitter, Facebook, Google News, and Instagram. For our newest video content, subscribe to our YouTube channel.
Read More: How to upgrade to Windows 10?
 and generative AI services.){kind=link}