ByteDance, the parent company of TikTok, has joined the ChatGPT trend along with Baidu and Alibaba Group Holding. They are currently testing a chatbot powered by large language models (LLM).
The project, codenamed “Grace,” is an experimental AI chatbot that is still in its early stages and is intended for internal testing purposes only, according to a representative from ByteDance.
About ByteDance’s LLM
The chatbot has been developed within Feishu, ByteDance’s enterprise collaboration platform known as Lark internationally. A tester who experienced it last month revealed that the testing has been limited to a select group of employees, and the information is not yet public.
This marks ByteDance’s second attempt to tap into the generative AI trend as companies worldwide aim to replicate the success of OpenAI’s ChatGPT, which gained significant attention after its launch last year. It’s important to note that ChatGPT is not officially accessible in China.
In May, ByteDance announced that it was in the early stages of exploring a chatbot called Tako. This AI-powered tool was designed to assist with content search on TikTok, and it was initially tested with select users in the Philippines.
ByteDance’s Algorithmic Prowess
The success of TikTok and its domestic counterpart Douyin can be largely attributed to their popular algorithmic recommendations. These recommendations guide users towards related content based on their preferences and activities.
ByteDance has been a strong advocate for AI since its establishment in 2012. AI technology forms the foundation of the algorithms used to deliver curated content to users across all of its flagship products, including the news aggregator Jinri Toutiao. This AI-powered recommendation system is often referred to as ByteDance’s secret sauce.
Furthermore, ByteDance employs AI techniques for video optimization, developing filters and special effects to enhance the entertainment value of short videos.
China’s big movements in generative AI
Baidu was the first major Chinese internet company to introduce a ChatGPT-like service called Ernie Bot in March. Alibaba, also launched its own alternative called Tongyi Qianwen in April. Other companies, such as AI developer SenseTime and voice recognition developer iFlyTek, have also entered the field.
AI applications in China require strict security assessments from regulatory bodies, including ChatGPT-like bots, before they can be offered as public services. The Cyberspace Administration of China (CAC) recently released a draft law stating that national internet regulatory departments must receive a security assessment application before these AI products can operate for public use.
This regulation has raised concerns and created uncertainty for large Chinese tech companies, including Baidu Inc., which have unveiled ChatGPT-like bots and similar AI services. The new law introduces a rigorous process that these companies must comply with before their generative AI products can be made available to the public.