Home > News content

Employees sold shares ahead of time, Baidu hastily launched the Chinese version of ChatGPT, and the inside story was exposed.

via:凤凰科技     time:2023/3/10 9:04:34     readed:333

According to news from March 10, Beijing time, Baidu plans to hold a press conference on March 16 to announce its ChatGPT competition: "Wen Xin Yi Yan".

百度

Be in a hurry to catch up

Hundreds of people on Baidu are working around the clock in order to develop an artificial intelligence chat robot called Wen Xin Yinyan, according to people familiar with the matter. In addition, Baidu has transferred staff from other teams to borrow powerful computer chips.

Some people familiar with the matter said that the artificial intelligence model, which is the basis of chatbots, was still undergoing data training before the scheduled release date of Wen Xin Yi Yan. Some employees say they don't have enough time to develop a well-functioning product. The release of Wen Xin Yi Yan is regarded as a highly anticipated event in China's science and technology industry.

Baidu Meter launched this product in stages, first opening public testing to a limited number of users. Baidu will first integrate Wen Xin Yiyan into its search engine last month and will open it to the public in March.

The haste reflects Baidu's purposeful gamble to stay ahead of Chinese competitors that recently announced similar plans. Baidu said it had signed agreements with more than 400 Chinese companies that would be able to use Wenxin in their products and services. Baidu said that in return, chatbots will gain experience running in different scenarios to help improve their performance.

百度CEO李彦宏

Baidu has lost the favor of investors in recent years, and the successful release of Wenxin Yiyan may help it return to the ranks of China's best technology companies. But failure could cause it to suffer a similar fate to Google. Until recently, Google insisted on taking a relatively cautious approach to artificial intelligence and made a factual error in its artificial intelligence chat robot demonstration, resulting in a market capitalization loss of $100 billion.

Multiple Challen

Over the years, Baidu has invested heavily in developing large language models. The large language model is the foundation of ChatGPT, and it adapts Google and ChatGPT English versions into Chinese. Baidu released its first language model in 2019, named Wenxin. Previously, Google named its model "BERT", and both cited characters from the cartoon Sesame Street.

At the end of December last year, as ChatGPT became more and more popular, Baidu CEO Robin Li briefed employees on their new progress in the field of chat robots. "We have such cool technology, but can we turn it into a product that everyone needs?" He said, "this is actually the most difficult step, but it is also the greatest and most influential step."

In early January, Baidu executives told its natural language processing team to start using the Wenxin language model to build a product similar to ChatGPT, according to people familiar with the matter. However, the project faces challenges, many of which have been shared by other artificial intelligence developers working on ChatGPT-like technologies.

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

百度依靠英伟达芯片训练“文心一言”

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

Baidu relies on Nvidia chip to train "Wen Xin Yi Yan".

translate engine: Tencent

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments