"OpenAI also shaken"... Is the AI industry shifting to 'open-source'?
OpenAI has officially suggested the possibility of shifting its "AI model development strategy," signaling a potential shift in the AI industry, which has traditionally adhered to the development of closed AI models. According to the information technology (IT) industry on the 3rd, Sam Altman, CEO of OpenAI, recently mentioned during a Reddit "Ask Me Anything" (AMA) session, "We were on the wrong side of history regarding open source" and added, "Personally, I believe there is a need to create a different form of open-source strategy." Kevin Weil, OpenAI's Chief Product Officer (CPO), also stated, "We are considering the possibility of open-sourcing older models," officially confirming OpenAI's shift away from its previous closed-source strategy. In fact, on the 31st (local time), OpenAI released a lightweight version of its inference AI model, 'o3 Mini,' for free. The 'o3 Mini' offers excellent performance in the fields of mathematics, science, and coding, boasting faster response times and higher cost-efficiency than its predecessor, the 'o1.' According to OpenAI, despite its low costs—$0.55 per million input tokens and $4.40 per million output tokens—the 'o3 Mini' maintains top-tier inference capabilities. This move is widely believed to be influenced by the Chinese AI startup DeepSeek. Recently, DeepSeek made a significant impact on the AI industry by unveiling its ultra-low-cost, high-performance AI model, 'R1.' The model's ability to achieve GPT-4 level performance at a fraction of the cost has introduced a new paradigm for AI model development. In fact, the AI model 'R1' developed by DeepSeek has been praised by executives from major U.S. tech companies such as OpenAI, Microsoft, and Apple for its 'low-cost, high-efficiency' innovation. Even Silicon Valley's prominent venture capitalist, Marc Andreessen, called it "one of the most amazing and impressive innovations I have ever seen" and referred to it as the "Sputnik moment" of the AI field. The term "Sputnik moment" refers to the moment when a leading nation in technology is shocked by an unexpected challenge from a latecomer, originating from the 1957 event when the Soviet Union launched the satellite 'Sputnik 1' before the United States. However, Sam Altman, CEO of OpenAI, mentioned in an online meeting last year that "the o3 Mini will be released by the end of January 2025, and soon after, the o3 will also be made public." Therefore, this release was part of the scheduled plan, and DeepSeek's emergence may not be the direct cause. Nonetheless, there are analyses in the global AI industry suggesting that the launch of o3 Mini is closely related to the "DeepSeek shock." U.S. IT media outlet TechCrunch analyzed, "OpenAI's decision to expand access to high-end AI models while securing a competitive edge over DeepSeek reflects a strategic move." Until now, companies like OpenAI and Google have dominated the market based on closed AI models. They developed AI models by investing astronomical capital and leveraging exclusive data. However, challengers like DeepSeek and Meta have been introducing open-source models, quickly improving AI performance and expanding their influence in the market. As a result, the future of the AI industry is expected to be significantly shaped by the competition between open-source and closed models. The industry anticipates that if the efficiency of open-source models is proven, even latecomers could develop high-performance AI with relatively lower costs. On the other hand, if closed models continue to maintain dominance, there is concern that the AI market could become entrenched in a structure dominated by a few large corporations. ChatGPT를 사용하여 번역한 기사입니다.
