share
 

DeepSeek's AI breakthrough widely recognized by US tech industry

0 Comment(s)Print E-mail Xinhua, February 11, 2025
Adjust font size:

A boy tries an AI-driven pad for learning assistance at the 7th World Voice Expo in Hefei, east China's Anhui Province, Oct. 24, 2024. (Xinhua/Fu Tian)

With its recent AI model breakthrough, Chinese artificial intelligence (AI) company DeepSeek has drawn a flurry of positive reactions from leading U.S. tech firms.

The company's R1 reasoning model, released last month, has been widely compared with OpenAI's currently most advanced model o1. But the R1 model was built at a fraction of what major U.S. AI labs spent on computing power.

DeepSeek's breakthrough in efficiency has received widespread acclaim from the U.S. tech industry, including its AI counterparts and leaders of tech giants.

Microsoft CEO Satya Nadella acknowledged DeepSeek's "genuine innovations" and the company has integrated the R1 model into Microsoft's developer platforms, Azure and GitHub. Nadella stressed the importance of taking developments from China seriously, citing the remarkable efficiency of DeepSeek's open-source model.

Meta CEO Mark Zuckerberg also highlighted the benefits of DeepSeek's published innovations.

"DeepSeek had a few pretty novel infrastructure optimization advances, which, fortunately, they published. We can not only observe what they did but read about it and implement it, so that'll benefit us," Zuckerberg said at a recent company meeting.

Nvidia, despite initial stock market concerns in response to DeepSeek's release, praised the R1 model as "an excellent AI advancement," describing the company's approach as a prime example of test-time scaling -- one of three key scaling methods currently shaping AI development.

Google CEO Sundar Pichai joined the chorus of praise, acknowledging DeepSeek's "very, very good work" and suggesting that lowering AI costs benefits both Google and the broader AI industry.

Like Microsoft, Amazon has embraced the new technology by allowing developers to leverage the R1 model through Amazon Web Services, describing it as "powerful and cost-effective."

DeepSeek's breakthrough has also impressed its U.S. counterparts like OpenAI and Perplexity.

OpenAI CEO Sam Altman described DeepSeek's R1 model as "impressive," particularly in its performance relative to cost. In response to this new competition, Altman announced that OpenAI would accelerate the release of improved models.

Former OpenAI executive Zack Kass reinforced the positive sentiment, calling the R1 model a breakthrough. "Being excited about progress in science is something that we should all want, and seeing the cost of a critical resource come down is also something we should want," Kass told Yahoo Finance.

Perplexity CEO Aravind Srinivas also lauded DeepSeek's AI model, emphasizing that the company is not simply copying existing technology but innovating in significant ways.

DeepSeek's ability to create efficient solutions marks a significant milestone in AI development, said Srinivas. "Because DeepSeek had to find a way to get around various limitations, it actually created something more efficient. They came up with many clever solutions."

Srinivas also pointed out that some of DeepSeek's innovations are so impressive that other major companies might adopt them.

The release of the R1 model and the publication of DeepSeek's methods have sparked what many see as a potential paradigm shift in the AI industry. With the knowledge of how to create powerful reasoning models now in the public domain, experts anticipate a surge of free, highly capable AI models in the near future.

An analysis by consulting firm KPMG suggests that DeepSeek's emergence could reshape the industry through several key factors. The analysis noted that the company's performance rivals advanced closed-source models, while its cost-efficiency and open-source approach enable developers and researchers worldwide to learn from and build upon its work.

The MIT Technology Review reported that DeepSeek's innovations demonstrate that reasoning models are less complicated to build than previously thought. DeepSeek's reinforcement learning techniques, which often eliminate the need for human feedback, were cited as a significant factor in reducing development costs.

Industry insiders said this development could level the playing field between large tech companies and smaller startups, potentially fostering more collaboration and innovation in the AI sector.

"This could be a monumental moment," Itamar Friedman, CEO of AI coding startup Qodo, was cited as saying in the report. 

Follow China.org.cn on Twitter and Facebook to join the conversation.
ChinaNews App Download
Print E-mail Bookmark and Share

Go to Forum >>0 Comment(s)

No comments.

Add your comments...

  • User Name Required
  • Your Comment
  • Enter the words you see:   
    Racist, abusive and off-topic comments may be removed by the moderator.
Send your storiesGet more from China.org.cnMobileRSSNewsletter
主站蜘蛛池模板: 激情射精爆插热吻无码视频| 高清欧美性猛交xxxx黑人猛交| 少妇大叫太大太爽受不了| 久久国产精品-国产精品| 欧美乱大交xxxxx另类| 亚洲欧美日韩精品久久奇米色影视 | 中文字幕色网站| 国产精品香蕉在线观看| 99在线精品免费视频| 女人扒下裤让男人桶到爽| 三上悠亚破解版| 成年午夜无码av片在线观看| 久久久婷婷五月亚洲97号色| 日韩精品无码一区二区视频| 亚洲一区二区三区在线观看网站| 欧美日韩国产一区二区三区欧| 国语自产偷拍精品视频偷拍| 一本精品99久久精品77| 手机看片中文字幕| 久久99精品国产自在现线小黄鸭| 日本特黄特黄刺激大片| 久久综合九色综合精品| 真实乱视频国产免费观看| 啊轻点灬大ji巴太粗太长h| 草莓视频成人在线观看| 国产亚洲综合久久| 韩国18福利视频免费观看| 国产成人啪精品| 成年人网站免费视频| 国产日韩一区二区三区在线播放| mm131美女爱做视频在线看| 性xxxx黑人与亚洲| 两个人一起差差差30分| 成人午夜精品无码区久久| 中文字幕在线播放视频 | 蜜桃视频在线观看官网| 国产凸凹视频一区二区| 青青草国产免费久久久下载| 国产国语在线播放视频| 韩国三级大全久久网站| 国产午夜av秒播在线观看|