DeepSeek ignites the 'deep seek' for humanity – Democratizing AI
DeepSeek motivates the quest for a deeper understanding of humanity through the democratization of AI.
TroibNews 
**Editor's note:** With the launch of DeepSeek's open-source AI model, the barriers to entry and the costs associated with AI development have seen a significant reduction. It is expected that AI's influence in human society will continue to grow, impacting the labor market and integrating more deeply into everyday life. As the annual Two Sessions get underway, CN's SCI-Tech Section highlights critical issues related to artificial intelligence and initiates the "DeepSeek inspires the 'deep seek' for humanity" series. This series, written by Li Mingling, an assistant researcher at the Institute of Public Policy at South China University of Technology, delves into vital questions regarding how to ensure that technological progress upholds fairness, justice, and human welfare without undermining human value. The viewpoints expressed in the article are those of the author and do not necessarily reflect those of CN.
The domestic Large Language Model DeepSeek has recently captivated the internet, prompting many in China to recognize that the era of AI has indeed arrived.
DeepSeek's remarkable ability to deliver performance on par with advanced LLMs like ChatGPT-4o but at a much lower cost has ignited discussions across the AI community and beyond. Although China had already introduced dozens of prominent Large Language Models before DeepSeek, these models faced challenges in achieving widespread adoption, with AI primarily being a topic of interest for tech enthusiasts, the younger population, or those with higher education. DeepSeek's emergence has made the topic of AI resonate with a far broader audience, including less formally educated individuals and older demographics, such as those aged 50-60. In this way, DeepSeek has been instrumental in bringing the discussion of AI into mainstream awareness among the general public in China.
While numerous articles have examined the global competition, innovation mechanisms, technological environment, and corporate cultures driving AI advancement, there remains an essential question for us to grapple with in the coming years: how should we navigate and adapt to the AI era?
This inquiry extends beyond AI research and development; it pertains to the direction AI evolution takes, the role of humanity in this new era, and the societal transformations that AI will enact—questions integral to our future.
Before the arrival of DeepSeek, a distinct digital divide existed, characterized by unequal access, usage, and benefits related to AI among various groups.
DeepSeek has been pivotal in bridging this gap. Beyond raising awareness of LLMs, it has encouraged greater participation from developers in AI innovation and made AI technologies more accessible and affordable for a wider audience. AI appears set to evolve into a public good that transcends social class, educational backgrounds, and national borders. Today, DeepSeek has expedited the realization of that vision.
In this context, the research and development team's perspective of AI as a public good is crucial to DeepSeek's impact. The model promotes digital equality in AI through two essential avenues: development and usage.
Firstly, through open-source initiatives, DeepSeek has altered the development landscape from a structure dominated by a few leading firms to one that fosters broad participation from numerous developers.
Open-source entails making a product's source code publicly available through relevant platforms, enabling others to access, modify, distribute, and contribute to the software. On December 26, 2024, DeepSeek launched the V3 Model, accompanied by a 53-page technical report, marking it as the world's first fully open-source Model of Mixed Expertise. Subsequently, on January 20, 2025, DeepSeek introduced the R1 Model and published its source code on Hugging Face, an open-source community platform.
In contrast, earlier major players like OpenAI and Google predominantly focused on closed-source models for their most sophisticated AI systems, creating algorithmic "black boxes" with limited transparency for external researchers and developers. DeepSeek’s open-source model can be seen as a movement towards democratizing AI. This approach not only enables users to run models locally with their databases but also empowers developers to make tailored modifications and engage in secondary development, enhancing the model's optimization.
While most researchers may be viewed as grassroots developers compared to leading tech companies, this does not preclude them from making groundbreaking advancements. According to a recently published CNBC report, Yann LeCun, chief AI scientist at Meta, stated that DeepSeek's success illustrates a victory for open-source AI models rather than a triumph for China over the United States.
Secondly, DeepSeek changes the usage model of AI from one restricted to a privileged few to one that is broadly available to the public, employing a strategy of low or even free pricing.
DeepSeek-R1's Application Programming Interface service costs 30 times less than ChatGPT-o1. Additionally, ordinary users can not only receive responses to a variety of inquiries on DeepSeek but can also observe the deep thinking and reasoning involved in those answers. In contrast, accessing this advanced feature on ChatGPT-o1 requires users to obtain a membership, which can be as high as $200 per month.
DeepSeek has directly disrupted the existing profit models of certain tech companies and significantly hastened the adoption of inclusive AI applications. Recently, prompted by DeepSeek's success, OpenAI announced an open-source release and low-pricing strategy for its o3-mini Model. Baidu's ERNIE also revealed plans for free access starting April 1. From the standpoint of technology diffusion, the democratization of AI is rapidly gaining traction.
The advancements in algorithms, enhanced computing power efficiency, and open-source initiatives have constituted the technological bedrock allowing AI to reach the general populace. This democratization is set to herald an era of more meaningful technological competition. As Jeffrey Ding, Assistant Professor at George Washington University, elucidates in his book *Technology and the Rise of Great Powers*, great power competition hinges not merely on who leads in innovation but on who can effectively adopt and diffuse innovations across diverse industries.
Historically, throughout industrial revolutions—from the steam engine and electricity to the advent of information networks—true progress has belonged to those who most adeptly encouraged the dissemination of technology.
Camille Lefevre contributed to this report for TROIB News
The domestic Large Language Model DeepSeek has recently captivated the internet, prompting many in China to recognize that the era of AI has indeed arrived.
DeepSeek's remarkable ability to deliver performance on par with advanced LLMs like ChatGPT-4o but at a much lower cost has ignited discussions across the AI community and beyond. Although China had already introduced dozens of prominent Large Language Models before DeepSeek, these models faced challenges in achieving widespread adoption, with AI primarily being a topic of interest for tech enthusiasts, the younger population, or those with higher education. DeepSeek's emergence has made the topic of AI resonate with a far broader audience, including less formally educated individuals and older demographics, such as those aged 50-60. In this way, DeepSeek has been instrumental in bringing the discussion of AI into mainstream awareness among the general public in China.
While numerous articles have examined the global competition, innovation mechanisms, technological environment, and corporate cultures driving AI advancement, there remains an essential question for us to grapple with in the coming years: how should we navigate and adapt to the AI era?
This inquiry extends beyond AI research and development; it pertains to the direction AI evolution takes, the role of humanity in this new era, and the societal transformations that AI will enact—questions integral to our future.
Before the arrival of DeepSeek, a distinct digital divide existed, characterized by unequal access, usage, and benefits related to AI among various groups.
DeepSeek has been pivotal in bridging this gap. Beyond raising awareness of LLMs, it has encouraged greater participation from developers in AI innovation and made AI technologies more accessible and affordable for a wider audience. AI appears set to evolve into a public good that transcends social class, educational backgrounds, and national borders. Today, DeepSeek has expedited the realization of that vision.
In this context, the research and development team's perspective of AI as a public good is crucial to DeepSeek's impact. The model promotes digital equality in AI through two essential avenues: development and usage.
Firstly, through open-source initiatives, DeepSeek has altered the development landscape from a structure dominated by a few leading firms to one that fosters broad participation from numerous developers.
Open-source entails making a product's source code publicly available through relevant platforms, enabling others to access, modify, distribute, and contribute to the software. On December 26, 2024, DeepSeek launched the V3 Model, accompanied by a 53-page technical report, marking it as the world's first fully open-source Model of Mixed Expertise. Subsequently, on January 20, 2025, DeepSeek introduced the R1 Model and published its source code on Hugging Face, an open-source community platform.
In contrast, earlier major players like OpenAI and Google predominantly focused on closed-source models for their most sophisticated AI systems, creating algorithmic "black boxes" with limited transparency for external researchers and developers. DeepSeek’s open-source model can be seen as a movement towards democratizing AI. This approach not only enables users to run models locally with their databases but also empowers developers to make tailored modifications and engage in secondary development, enhancing the model's optimization.
While most researchers may be viewed as grassroots developers compared to leading tech companies, this does not preclude them from making groundbreaking advancements. According to a recently published CNBC report, Yann LeCun, chief AI scientist at Meta, stated that DeepSeek's success illustrates a victory for open-source AI models rather than a triumph for China over the United States.
Secondly, DeepSeek changes the usage model of AI from one restricted to a privileged few to one that is broadly available to the public, employing a strategy of low or even free pricing.
DeepSeek-R1's Application Programming Interface service costs 30 times less than ChatGPT-o1. Additionally, ordinary users can not only receive responses to a variety of inquiries on DeepSeek but can also observe the deep thinking and reasoning involved in those answers. In contrast, accessing this advanced feature on ChatGPT-o1 requires users to obtain a membership, which can be as high as $200 per month.
DeepSeek has directly disrupted the existing profit models of certain tech companies and significantly hastened the adoption of inclusive AI applications. Recently, prompted by DeepSeek's success, OpenAI announced an open-source release and low-pricing strategy for its o3-mini Model. Baidu's ERNIE also revealed plans for free access starting April 1. From the standpoint of technology diffusion, the democratization of AI is rapidly gaining traction.
The advancements in algorithms, enhanced computing power efficiency, and open-source initiatives have constituted the technological bedrock allowing AI to reach the general populace. This democratization is set to herald an era of more meaningful technological competition. As Jeffrey Ding, Assistant Professor at George Washington University, elucidates in his book *Technology and the Rise of Great Powers*, great power competition hinges not merely on who leads in innovation but on who can effectively adopt and diffuse innovations across diverse industries.
Historically, throughout industrial revolutions—from the steam engine and electricity to the advent of information networks—true progress has belonged to those who most adeptly encouraged the dissemination of technology.
Camille Lefevre contributed to this report for TROIB News
