TL;DR:
- Baidu’s Ernie Bot emerged as the top-ranked ChatGPT alternative among Chinese tech companies in an evaluation conducted by Xinhua Institute.
- However, Ernie Bot still lags behind OpenAI’s GPT-4 and GPT-3.5, which are not available in mainland China.
- Xinhua’s test assessed language skills, intelligence, and the chatbot’s ability to enhance professional productivity.
- Chinese technology firms are striving to compete with ChatGPT, driven by renewed interest in AI and LLMs.
- Clue’s evaluation ranked Qihoo 360’s Smart Brain as the top Chinese model, followed by iFlytek’s SparkDesk.
- Local competitors face challenges such as training model gaps, data availability, and the uniqueness of the Chinese language.
- Baidu’s founder, Robin Li Yanhong, acknowledged Ernie Bot’s lag but emphasized its potential for rapid evolution with human feedback.
- iFlytek’s chairman, Liu Qingfeng, expressed confidence that their chatbot would catch up with ChatGPT by October.
Main AI News:
In the race for supremacy among large language models (LLMs), Baidu’s Ernie Bot has emerged as the leader among the ChatGPT alternatives offered by Chinese tech companies. According to a recent evaluation conducted by Xinhua Institute, a think tank affiliated with the Xinhua state news agency, Ernie Bot outperformed competitors such as Alibaba Group Holding’s Tongyi Qianwen, iFlytek’s SparkDesk, and SenseTime’s SenseChat.
Despite its success, Ernie Bot still falls behind OpenAI’s GPT-4 and GPT-3.5, which are not officially available in mainland China. It is worth noting that GPT-3.5 powers the original version of ChatGPT, which was publicly released in November of last year.
Xinhua’s comprehensive test assessed various factors, including basic language skills, such as prompt comprehension and response generation, as well as intelligence encompassing common sense, logical reasoning, and subject knowledge in fields such as mathematics, physics, finance, and literature. The test also evaluated the chatbots’ ability to enhance the productivity of professionals, ranging from journalists and painters to designers, marketing specialists, lawyers, and researchers.
Following the success of ChatGPT, Chinese technology firms have been quick to introduce their own chatbot offerings, reigniting interest in the potential of artificial intelligence (AI), particularly in the domain of LLMs—deep-learning algorithms capable of understanding and generating text, images, and videos.
Another LLM evaluation conducted by Clue, a Chinese website tracking AI research development, ranked Qihoo 360’s Smart Brain as the top-performing Chinese model, closely followed by iFlytek’s SparkDesk. However, both models still fell short of GPT-4, GPT-3.5, and US start-up Anthropic’s Claude.
Xinhua Institute noted in its report that the test was subject to time and conditional constraints, introducing a certain level of subjectivity to the results.
Despite the progress made by Chinese tech companies, some experts believe that foreign companies are likely to maintain their lead in cutting-edge AI technology on the global stage. Establishing local competitors to ChatGPT poses challenges, including gaps in training models, the availability of data sets, and the unique nature of the Chinese language itself.
During an interview with Chinese tech news outlet 36Kr in late March, Baidu’s founder and CEO, Robin Li Yanhong, mentioned that Ernie Bot was “two months behind” ChatGPT. However, he later clarified that it could take Baidu an indefinite amount of time, ranging from a short period to potentially “forever,” to close that two-month gap. Li also expressed that Ernie Bot, as the first ChatGPT-like service offered by a major Chinese tech firm, has the potential to evolve rapidly with human feedback.
At the launch event of SparkDesk in early May, Liu Qingfeng, founder and chairman of iFlyTek, confidently stated that their chatbot would catch up with ChatGPT by the end of October, indicating the determination of Chinese tech companies to bridge the AI gap.
Conclusion:
Baidu’s Ernie Bot’s success in leading the Chinese LLM rankings highlights the progress made by Chinese tech companies in the AI domain. However, the continued superiority of OpenAI’s GPT models suggests that foreign companies are likely to maintain an edge in cutting-edge AI technology globally. Establishing competitive local alternatives to ChatGPT poses challenges, including training model gaps and the distinctive characteristics of the Chinese language. Despite these hurdles, Chinese tech companies remain determined to bridge the AI gap and close the distance with their international counterparts.
