原文链接：

https://digitalpaper.stdaily.com/http_www.kjrb.com/ywtk/html/2025-03/01/content_585318.htm?div=-1

原文如下：

The AI industry has undergone transformative changes over the past decade, with advances in machine learning, natural language processing, and computer vision reshaping the way we interact with technology.

Among the key players driving this revolution is Chinese-based DeepSeek, a free AI-powered chatbot and assistant that is pushing the boundaries of AI innovation. By developing cutting-edge solutions and fostering a culture of research and collaboration, DeepSeek has become a game-changer in the AI landscape.

A revolution with limited resources

Since the debut of ChatGPT in November 2022, the AI industry has been booming, with many tech giants like Meta, Microsoft, and Google investing billions of USD in the field and developing a range of similar models. AI has become an integral part of our daily lives with versatile applications that not only improve productivity, efficiency, and quality, but also reduce time and costs.

Such AI models require high computing power with many graphic processing units (GPUs). For example, U.S. multinational NVIDIA, one of the pioneering chip companies, focuses on advanced GPUs with high computing power, which is crucial for AI models. As a result, the company became the most valuable company in the world, with a stock value of 3.5 trillion USD.

However, in parallel with this AI revolution, the U.S. has restricted China from obtaining advanced AI chips, even via other countries, to limit the development of AI models, citing concerns that China could use AI technology for military purposes. As a result, many Chinese companies were restricted to downstream applications and linked AI to hardware systems.

After the first release of the AI open-source chatbot model on January 20, a relatively unknown company, DeepSeek, became the talk of the town, especially in Silicon Valley. The DeepSeek-R1 model competes with the leading OpenAI o1 model on the basis of capability, cost and speed.

The company has focused on software-based development and is trying to find an alternative way, rather than focusing on advanced hardware systems and supercomputers, to develop a robust AI model using limited resources, including AI chips and GPUs. It has used a chain of thoughts, reinforcement learning, trial and error, reward engineering, distillation, emergent behavior network, mixing of experts, and memory-efficient techniques to circumvent the scarcity of AI chips and redesign the radical structure of AI models.

The miraculous rise of DeepSeek, with low-cost and memory-efficient techniques, not only stunned the tech giants, but also led to a rethinking about the cost of AI training and its relevant use of AI chips. As a result, NVIDIA lost 600 billion USD in market value in one night, and many tech giants, such as ByteDance and Alibaba, lowered the price of their AI models. DeepSeek has shown how to leverage memory and computing power to train and run AI models with billions of parameters, paving the way for an AI industry revolution with limited resources.

AI access for all

One of DeepSeek's most significant contributions to the AI industry is its commitment to democratizing access to AI technologies. Recognizing that the benefits of AI should be available to all, the company has developed easy-to-use models, tools, and platforms that enable users to leverage AI without requiring extensive technical expertise. Soon after its release, the DeepSeek App became the most downloaded free App in the U.S. Apple Store, recognizing its popularity made DeepSeek available to its users.

Unlike ChatGPT, DeepSeek is an open-source AI model under the license of the Massachusetts Institute of Technology, which means that anyone can use the model for professional or personal purposes without restrictions on tokens and parameters. In addition, anyone can run the DeepSeek-R1 model on a local computer based on their hardware configuration, making the model accessible to a wide range of users and more extensible to link with versatile applications and learning.

Through its cutting-edge technologies, low cost, and commitment to democratizing access to AI, the company is redefining what is possible with AI. DeepSeek is paving the way for a future where AI is not only powerful, but also inclusive, responsible, and beneficial to all.

In a world increasingly shaped by technology, DeepSeek stands as a beacon of progress, demonstrating how AI can be a force for good and a catalyst for positive change. With its impressive track record and ambitious vision, DeepSeek is not just transforming the AI industry — it is shaping the future of humanity.

Dr. Md Altab Hossin is a Bangladeshi expert at the School of Innovation and Entrepreneurship, Chengdu University.

中文翻译：

DeepSeek使AI访问民主化

随着机器学习、自然语言处理和计算机视觉的进步对我们与技术的互动方式的重塑，人工智能产业在过去十年经历了变革。

　　推动这场革命的关键之一是总部位于中国的DeepSeek，一个正在推动人工智能创新边界的免费人工智能聊天机器人和助手。通过不断开发前沿问题解决方案并培养一种研究与合作的文化，DeepSeek已然成为了人工智能领域的游戏规则改变者。

资源有限的革命

自2022年11月ChatGPT首次亮相以来，人工智能行业一直在蓬勃发展，Meta、微软和谷歌等许多科技巨头在该领域投资了数十亿美元并开发了一系列与其相似的模型。人工智能应用功能众多，不仅可以提高生产率、效率和质量，还可以减少时间和成本，并使人工智能成为了我们日常生活中不可或缺的一部分。

这样的人工智能模型需要具有许多图形处理单元（GPU）的高计算能力。例如，美国跨国公司NVIDIA，一家世界领先的芯片公司，专注于具有高计算能力的，对于AI模型至关重要的先进GPU。因此，该公司斩获了高达3.5万亿美元的股票价值并成为了世界上价值最高的公司。

然而，在这场人工智能革命的同时，美国声称中国可能将人工智能技术用于军事目的并开始限制中国甚至从美国以外的国家对先进人工智能芯片的获取，以限制人工智能模型的发展。因此，许多中国公司受限于下游应用，并将人工智能连接到硬件系统之上。

　　自从人工智能开源聊天机器人模型于1月20日首次发布后，一家相对不知名的公司DeepSeek成为了人们话题的焦点，尤其是在硅谷。DeepSeek-R1模型在性能、成本和速度方面皆可与领先的OpenAI o1模型相媲美。

该公司一直专注于基于软件的开发，并试图找到一种替代专注于先进的硬件系统和超级计算机的方法，而非以利用有限的资源（包括AI芯片和GPU）开发一个强大的AI模型。其采用了一连串思想、强化学习、试错、奖励工程、数据净化、紧急行为网络、专家混合和高效记忆技术来规避人工智能芯片的稀缺性，并重新设计人工智能模型的基本结构。

　　DeepSeek凭借低成本和高效存储技术的奇迹般崛起，不仅震惊了科技巨头，还引发了对人工智能培训成本及其相关人工智能芯片使用的重新思考。结果，NVIDIA一夜之间就损失了6000亿美元的市场价值，许多科技巨头，如ByteDance和阿里巴巴，都降低了他们的人工智能模型的价格。DeepSeek展示了如何充分利用内存和计算能力来训练和运行具有数十亿参数的人工智能模型，并为有限资源的人工智能行业革命开创了先河。

人人享有的人工智能

DeepSeek对人工智能行业最重要的贡献之一是其对实现人工智能技术的民主化的决意。认识到AI带来的好处应该为所有人所用，该公司开发了易于使用的模型、工具和平台，使即便缺乏技术专业知识的用户也能够利用AI。发布后不久，DeepSeek应用成为美国苹果商店下载量最多的免费应用程序，认识到它的受欢迎程度，使DeepSeek对其用户可用。

　　与ChatGPT不同，DeepSeek是麻省理工学院许可下的开源AI模型，这意味着任何人都可以将该模型用于专业或个人用途，且不受分词和参数的限制。此外，任何人都可以根据其硬件配置在本地计算机上运行DeepSeek-R1模型，从而使该模型可供广泛的用户访问，并具有更大的可扩展性，以链接到多功能应用程序和学习。

　　通过其尖端技术、低成本以及对人工智能访问民主化的决意，该公司正在重新定义人工智能的可能性。DeepSeek正在为一个人工智能不仅强大，而且包容、负责且惠及所有人的未来开创先河。

　　在一个越来越受技术影响的世界里，DeepSeek作为一个进步的里程碑，展示了人工智能如何成为一股向善的力量和积极变革的催化剂。凭借其令人印象深刻的业绩和雄心勃勃的愿景，DeepSeek不仅在改变人工智能行业，还在塑造人类的未来。

　　Md Altab Hossin博士是成都大学创新创业学院孟加拉国籍专家。