Generative AI, the architecture that has delivered today’s large language models (LLMs) such as ChatGPT, can never be scaled up to create artificial general intelligence (AGI), an AI that matches or outperforms humans’ cognitive capabilities, according to 76 percent of the 475 AI scientists responding to a poll organized by Association for the Advancement of Artificial Intelligence.
Unlike AI Western developers, Chinese company DeepSeek has reacted to its recent fame and soaring sales by focusing even more heavily on research, not chasing greater and greater revenue.
New graduates from Harvard, Stanford, and other leading U.S. universities are applying for work at China’s Manycore AI company, Bloomberg reported.
On 20 January, Chinese start-up DeepSeek revolutionized the AI industry by unveiling a new, world-class platform that it had created for a fraction of the cost that Western industry giants had spent.
DeepSeek, the Chinese AI start-up that shocked the industry in January by claiming a world-class AI was built for a sliver of what Western developers have spent, now says its V3 and R1 models show a cost-profit ratio as high as 545 percent every day.
The kind of cheaper, less energy-intense AI created by Chinese start-up DeepSeek will not slow the headlong demand for AI data centers around the world, according to Schneider Electric, a major supplier of electrical infrastructure to the centers and other industrial sites.
The nonprofit Allen Institute for AI has announced its new open-source AI competes with or bests DeepSeek’s V3 model and OpenAI’s ChatGPT 4-o in several standard benchmark tests.
China’s tech stocks are surging in price in the wake of start-up DeepSeek’s announcement of its world-class artificial intelligence system developed at a fraction of the costs that Western developers have incurred to create theirs.
Share prices of the 50 largest stocks in the Standard & Poor’s 500 index are more fragile than at any time this century, Bank of America (BoA) analysts warn in a new report.
China’s DeepSeek AI was trained for $6 million using second-tier chips, the company says.