Home > Technology peripherals > AI > ChatGPT Zhihu topic views exceeded 200 million, OpenAI employees joined the discussion

ChatGPT Zhihu topic views exceeded 200 million, OpenAI employees joined the discussion

WBOY
Release: 2023-04-12 18:49:06
forward
1637 people have browsed it

ChatGPT Zhihu topic views exceeded 200 million, OpenAI employees joined the discussion

Recently, in the past three months, the number of discussions under the topic "ChatGPT" on Zhihu has reached 103,000, and the total number of views has exceeded 210 million, approaching the number under the topic "AlphaGo" since 2015. Total discussion views.

Including global artificial intelligence education leader Andrew Ng, founder of DeepLearning.AI, Chairman of Sinovation Ventures Kai-fu Lee, OpenAI employees and many other industry figures, experts and scholars joined the discussion.

In 2015, AlphaGo triggered the last round of heated discussions about AI, with a total of 210,000 discussions and 220 million views on Zhihu. Former Sogou CEO Wang Xiaochuan, Sinovation Ventures Chairman Kai-fu Li, Meta FAIR Research Institute researcher Tian Yuandong and other respondents answered questions on "What do you think of Google's Go AI AlphaGo defeating the European Go champion?" "Can AlphaGo defeat Lee Sedol?" "What do you think of Ke Jie's upcoming victory?" The human-machine battle with AlphaGo at the end of May? What impact will this have on Go, humans and artificial intelligence?" Discussions under such issues have attracted great attention.

The "ChatGPT" topic has accumulated 103,000 discussions in a short period of time. Although the number of discussions is less than half of AlphaGo, the number of views has reached 210 million, approaching the total number of discussion views on the AlphaGo topic since 2015. .

There is an interpretation from a public perspective. Under the question "What professions cannot be replaced by ChatGPT?", Kai-fu Lee, Chairman of Sinovation Ventures, summarized 10 kinds of "sit back and relax" jobs, 10 kinds of "near-miss" jobs, and proposed ways to save them in the post-ChatGPT era. General rules for your profession.

Zhang Junlin, head of new technology R&D at Sina Weibo, said "ChatGPT has confirmed the feasibility of model unification. What impact will this have on NLP practitioners in the next five years?" For NLP researchers, the outlook is very bleak."

Under the question "There are so many AI majors in China, why is there no ChatGPT in China?", Lan Zhenzhong, a former scientist at Google Artificial Intelligence Research Institute and now a distinguished researcher at West Lake University, brought his own prediction: "The training method of ChatGPT At present, teams with experience in large model research should know how to do it, but it requires more machines and longer time to train. And the benefits obtained from the training of large models will be less and less, which means that although OpenAI will invest more money to develop subsequent versions, but since the subsequent training will become increasingly difficult, we can keep up as long as we continue to follow suit."

OpenAI China Staff Zhihu Respondent Trinkle revealed that ChatGPT is not like outside rumors, with dozens of doctors training, "only a dozen core training staff." His main job is to "optimize the system during the entire reinforcement learning training to make it run faster and complete the reinforcement learning training as quickly as possible with the least GPU."

The above is the detailed content of ChatGPT Zhihu topic views exceeded 200 million, OpenAI employees joined the discussion. For more information, please follow other related articles on the PHP Chinese website!

Related labels:
source:51cto.com
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template