Since the Chinese artificial intelligence company Deepseek released an open version of its logical model R1 at the beginning of this week, many in the technology industry issued major statements about what the company made, and what this means to the state of artificial intelligence.
For example, the adventurous capitalist Mark Anderson. to publish Deepseek is “one of the most amazing and impressive discoveries that I have ever seen.”
R1 appears to match or outperform OPENAI’s O1 model. The company claims one of its models At a cost of only $ 5.6 million For training, compared to hundreds of millions of dollars paid by the leading American companies to train their employees.
It also appears to have achieved this in the face of US sanctions prohibiting the sale of advanced chips to Chinese companies. Mit Technology Review wrote The success of the company explains how the sanctions “push startups such as Deepseek to innovate in ways that give priority to efficiency, gathering resources, and cooperation.” (On the other side, The Wall Street Journal reported Liang Winfing from Deepseek recently told the Chinese Prime Minister that American export restrictions still form the bottleneck.)
CEO of Cori Neil Khosla I offered a simpler explanationClaiming that the company is a “CCP State Psyop”, which “pretends that the cost was low to justify the price determination low and hopes that everyone will turn to it (to harm the competitiveness of artificial intelligence in the United States.”) (A societal note was attached to his reference indicating that Khosla is No evidence of this is provided, and that his father Vinod is an investor in Openai.)
During, Journalist Holger Zschaepitz suggested Deepseek can represent the biggest threat to American stock markets – if a Chinese company can build an advanced model at a low cost, without reaching advanced chips, this would question the “benefit of hundreds of billions of dollars from technology.” Capital expenses are pumped into this industry.
In response, Gary Tan, CEO of Y Combinator Deepseek will actually be useful for its American competitors: “If the training models become faster and easier, the demand for reasoning (actual use in the real world of artificial intelligence) will grow and accelerate faster, ensuring the use of the supply of computing.”
And the head of artificial intelligence scholars in Mita Yan to be Argue Looking at the Deepseek advertisement through the lens of China against the United States. Instead, I suggest that the real lesson is that “open source models outperform the owned models.”
Lecun wrote: “Deepseek has benefited from open research and open sources (such as Pytorch and Llama from Meta).” “They reached new ideas and built them over the works of others. Since their works are published and open, everyone can benefit from them.
It seems that all this controversy pushes consumers to experience the product – from Sunday afternoon, Deepseek artificial intelligence assistant It is the best free app in the Apple App Store, and it applies directly to ChatGPT.
