Participate in the new and weekly newsletter for the latest updates and exclusive content on the leading AI coverage in the industry. learn more
Sam AltmanOpenai’s CEO, Impressive admission On Friday, his company may be “incorrectly aspects of history” regarding open source AI, and as competition from China intensifies and efficient open models gain traction, strategic earthquakes. It shows a change.
The frank approval came between Reddit.Please ask anythingSession, a few days after the Chinese AI company Deepseek Rattling global market The open source R1 model claims to be comparable to Openai systems at some costs.
“Yes, we are discussing [releasing model weights]” Altman wrote。 “We think we were on the wrong side of history, and I think we need to understand different open source strategies.” Not that it was not the highest priority of the company today.
The statement represents an amazing departure Openai’s increasingly unique approach In recent years, it has been criticized by some AI researchers and former allies. Elon MaskAppeal for those who have betrayed the original open source mission.
Deepseek’s Sam Altman: “We will maintain the lead”
Altman’s comment comes inside Market turmoil Triggered by the appearance of DeepSeek. Under claim that Chinese companies will build a JUST advanced AI model 5.6 million dollars training costs (However, the total development cost is so Probably much higher) Send NVIDIA stocks Sudden fallWe will wipe out the market value of nearly $ 600 billion. This is the biggest day for US companies in history.
“We create better models, but we don’t keep the lead more than the past few years,” Altman directly treated the effects of Deepseek at the same AMA.
Sam Altoman acknowledges that Openai’s closed strategy may be defective.
DeepSeek’s breakthrough emphasizes the changes in dynamics in AI development, regardless of whether a specific claim has been accurate. The company states that the results were achieved using only the results. 2,000 NVIDIA H800 GPU-Normally, less than the estimated 10,000 or more chips developed by major AI labs.
This approach suggests that the innovation of algorithms and the optimization of architecture are more important than raw computing. The revelation is built based on exclusive access to large -scale calculation resources, as well as Openai’s technical strategy.
Open source discussion: Innovation and security
However, the rise in DeepSeek has been strengthened. Concerns of national security。 The company preserves user data on a server on mainland China and may be subject to government access. Several US organizations are already moving to restrict their use, and NASA has become the latest in blocking the application by quoting security and privacy concerns.
The potential pivot to Openai’s open source marks the return to its roots. The company was established as a non -profit organization in 2015 and was assigned to bring interest in humanity to mankind. However, the transition to the “Capped profit” model and the closer approach are attracting criticisms from open source supporters.
“The correct reading is that” the open source model is beyond its own model, “said a meta -AI scientist. Yann Lecun of LinkedinWe support the appearance of DeepSeek. “They came up with new ideas and built them on other people’s work. Their jobs were published and open source, so everyone can make a profit. It’s open. It is the power of research and open source.
New chapter of AI development
Altman’s comment suggests that strategic changes may come, but he emphasized that open source is not the top priority on Openai. The pursuit of this reflects the complex reality facing the AI leader. It balances innovation, security and commercialization in the world of multi -polar AI.
Stakes is far beyond Openai’s revenue. Company decisions may be reconstructed by the entire AI ecosystem. Open -sourcing key models can accelerate innovation and democratize access, but may complicate the safety and security of AI.
The timing of entering Altman suggests that Openai may be reacting to the power of the market, not to lead them, not before that, not before the DeepSeek market shock. This reactive stance has a reversal of an impressive role for companies that have long positioned as an AI noise star.
One thing becomes clear from the debut of DeepSeek as the dust calms. The real confusion is not only technology or market value. It is to challenge the assumption that a closely protected AI model is the most certain path for artificial general information. From that point of view, Altman’s approval may be less about being in the wrong side of history, but rather recognizing that history itself has changed the course.
