Advertisement
Advertisement
Artificial intelligence
Get more with myNEWS
A personalised news feed of stories that matter to you
Learn more
Xu Li, CEO and co-founder of SenseTime, unveils SenseTime’s answer to ChatGPT, on April 10. Photo: Weibo

Explainer | Chinese tech firms are looking to create their own ChatGPT moment but do they match up to the US front runner?

  • Baidu was among the first Chinese hopefuls to get in on the act, unveiling its ChatGPT-challenger Ernie Bot in February
  • Alibaba has integrated its effort, Tongyi Qianwen, into its Slack-like online collaboration tool Dingtalk

Ever since Microsoft-backed OpenAI launched its ChatGPT product to critical acclaim last November, a growing list of Chinese tech firms have been rushing to push out competing technology for the domestic market, where the US-made generative artificial intelligence (AI) chatbot is not officially available.

Here is a run-through of some of the leading Chinese competitor products in this emerging technology hotspot:

Baidu

The Beijing-based search engine was among the first Chinese hopefuls to get in on the act, unveiling its ChatGPT-challenger Ernie Bot in February followed by a press event in mid-March.

Baidu founder Robin Li Yanhong showcases Ernie Bot, its answer to ChatGPT, at company headquarters in Beijing on March 16. Photo: Handout

Ernie Bot received mixed initial reviews, with some enthusiasts disappointed by the lack of a live demonstration at the launch event. The reception has subsequently improved, and Ernie Bot is now embedded within Baidu’s internal management platform with at least 90,000 business clients applying to try the service.

Baidu, which runs the world’s largest Chinese search engine, has in recent years poured considerable resources into various areas of AI, including natural language processing (NLP) and pre-trained large language models (LLMs), the technology behind ChatGPT’s meteoric rise.

Alibaba Group Holding

Alibaba, the owner of the South China Morning Post, has launched an AI chatbot dubbed Tongyi Qianwen, which means seeking truth from a thousand questions. It plans to integrate the service into all of the company’s product offerings, said CEO Daniel Zhang Yong at an April event.

The company’s LLM efforts fall under its research unit, the DAMO Academy, which focuses on frontier technologies.

Alibaba integrates own ChatGPT-style service to collaboration app DingTalk

This week, the Hangzhou-based company held an online event that gave a peek into how it has integrated Tongyi Qianwen into Dingtalk, a Slack-like online collaboration tool. One example of this is the content generation capacity powered by Tongyi, allowing Dingtalk to summarise previous conversations in a group chat when a new member is added to the discussion.

iFlyTek

iFlyTek and other Chinese AI unicorns such as Megvii and Cloudwalk, were early pioneers of AI in applications such as voice and facial recognition, and they are trying to catch up in generative AI.

iFlyTek says that its State Key Laboratory of Cognitive Intelligence, the only state-backed research facility of its kind, is poised to launch its 1+N Cognitive LLM early next month. According to the company, 1 denotes the foundational language and training model, on which N, which refers to a subset of models tailored for various industries, is being trained.

The company, which makes voice recorders and software to help its clients transcribe meeting or conversation recordings, has deep knowledge of NLP, which could be advantageous in its efforts to match ChatGPT.

SenseTime

One of the four Chinese AI “little dragons” trying to catch up with the generative AI trend, SenseTime in early April launched its SenseNova product. It is part of a set of models that cover key capabilities including computer vision, natural language processing and AI-generated content.

Its ChatGPT-like service, SenseChat, was also unveiled at the April event. It is able to understand complex prompts, or questions, and respond with seemingly believable answers.

Huawei Technologies Co

Telecoms equipment giant Huawei has also joined the ChatGPT fray with the early April launch of Pangu, named after a mythological figure who was believed by the Chinese people to have created the sky and the earth.

However, Pangu received an online backlash, with many netizens pointing to the lack of either a live or recorded demo when the product was announced.

The company expects to use its Pangu model across several industries including pharmaceuticals, gaming, meteorology and water utilities.

Qihoo 360

Chinese cybersecurity firm Qihoo 360, which has had a relatively low public profile in recent years, hit the headlines recently after the company’s founder and CEO Zhou Hongyi talked up the ChatGPT frenzy.

Qihoo 360’s founder shrugs off calls for pause in AI research

His company launched its 360 Smart Brain earlier this month, which has been made available for beta testing by corporate clients.

In a live demo of the Smart Brain, Zhou had to tweak questions multiple times on some topics to get the expected answer, indicating it is currently behind ChatGPT and Baidu’s Ernie Bot when it comes to one-shot accurate responses. “It feels like a kid”, said Zhou at the launch event.

Tencent Holdings

The Shenzhen-based internet and gaming giant has been relatively quiet on its current thinking around ChatGPT. In its latest earnings call, president Martin Lau said that “Tencent wanted to do it [generative AI] right rather than do it in a rush.”

However, there have been local media reports that Tencent has built a team to make HunyuanAide, a ChatGPT-like service based on its Hunyuan LLM. Similar to Alibaba, the reports say Tencent is looking to bake the service into several areas including WeChat, gaming, advertising and short video.

Chinese Academy of Sciences

The Institute of Automation at the Chinese Academy of Sciences was among the first Chinese academic facilities to launch an LLM. Back in 2021, way before ChatGPT set the global tech community on fire, the facility announced a multi-modal LLM called Zhidong Taichu. The facility hailed this LLM as the world’s first pre-trained model able to understand visual, audio and text prompts.

1