Baidu claims its new AI outperformed ChatGPT on certain metrics, according to China Science Daily.
China’s leading search engine provider, Baidu Inc., announced that its latest version of the Ernie AI model, Ernie 3.5, has surpassed the widely popular Microsoft-backed OpenAI chatbot on multiple significant measures. Baidu declared in a statement on Tuesday that Ernie 3.5 outperformed ChatGPT in comprehensive ability scores and surpassed GPT-4 in several Chinese capabilities. This claim was supported by a test conducted by the state newspaper China Science Daily, which utilised AGIEval and C-Eval datasets, commonly employed benchmarks for evaluating AI model performance.
OpenAI did not respond immediately to an email seeking comment, as it was outside of regular business hours.
Baidu’s announcement comes at a time of global excitement surrounding AI, largely sparked by the success of ChatGPT, which has now reached China and prompted numerous domestic companies to announce their own competitive products. The first significant Chinese tech company to release an AI product meant to compete with ChatGPT was Baidu, which unveiled its language AI, Ernie Bot, in March.
For the past three months, Ernie Bot has been undergoing invite-only testing. Ernie Bot is based on Baidu’s prior Ernie 3.0 AI model. Tencent Holdings and “Alibaba Group,” two other well-known Chinese IT behemoths, have also made their own AI models public.
The enhanced training and inference efficiency of the new model, according to Baidu, positions it for future versions that will be quicker and more cost-effective. The business also emphasised that the new approach permits external “plugins.”
These plugins are supplementary applications that enable Baidu’s AI to tackle more specific scenarios, such as summarising lengthy text and generating more precise answers. ChatGPT had already introduced plugin support in March, and now Baidu follows suit with its own capabilities.
About The Author
