麻省理工学院（MIT）的一项最新研究显示，大型语言模型（LLM）的能力增长速度远远超过著名的摩尔定律。MIT FutureTech的研究人员发现，LLM的能力大约每8个月就会翻一番。这一发现意味着人工智能技术的进步速度远远超过了计算机硬件性能提升的速度。
摩尔定律是英特尔联合创始人戈登·摩尔在1965年提出的，预测集成电路上的晶体管数量大约每两年会翻一番，从而使计算机的处理能力相应提升。然而，MIT的研究表明，LLM的能力提升主要依赖于算力，而这可能会在未来达到一个无法满足的算力需求点。
这一研究结果对于人工智能领域的未来发展具有重要意义，因为它揭示了人工智能技术的增长速度可能会受到算力限制的挑战。随着LLM的能力不断增强，对于算力的需求也将会持续增加，这可能会导致未来对高性能计算资源的需求远远超过当前的供应能力。
英文标题：MIT Study: AI’s Growth Rate Outpaces Moore’s Law
英文关键词：MIT Study, AI Growth, Moore’s Law
英文新闻内容：
A recent study by MIT has revealed that the growth rate of large language models (LLM) is surpassing Moore’s Law. Researchers from MIT FutureTech found that the capabilities of LLMs double approximately every eight months, far exceeding the rate of improvement in computer hardware performance predicted by Moore’s Law.
Moore’s Law, proposed by Intel co-founder Gordon Moore in 1965, states that the number of transistors on a microchip doubles approximately every two years, leading to a corresponding increase in computing power. However, the MIT study indicates that the growth of LLM capabilities is primarily driven by computational power, which may reach a point where the required computational resources cannot be met in the future.
The findings of this research have significant implications for the future of AI technology, as it suggests that the rapid growth of AI capabilities may eventually face challenges due to computational power limitations. As the capabilities of LLMs continue to expand, the demand for computational resources is likely to increase at a rate that exceeds the current and projected supply, potentially leading to a future where high-performance computing resources are in short supply.
【来源】https://mp.weixin.qq.com/s/HLHrhOkHxRPRQ3ttJLsfWA
Views: 2