What does open source AI mean, and why is DeepSeek different?

What does open source AI mean, and why is DeepSeek different?
  • DeepSeek has shaken up the AI ​​industry, surpassing ChatGPT to become the most downloaded free app in the U.S. Apple App Store.
  • What makes DeepSeek a key differentiator is that the Chinese app is open source, meaning anyone can copy, download and build on it.
  • The World Economic Forum’s AI for Industry Transformation initiative is exploring the challenges and opportunities of AI-driven innovation.

Has artificial intelligence had its own Sputnik moment? In other words, has the Chinese company DeepSeek created a cheaper, more energy-efficient, and easier-to-use way to develop AI models?

Well-known technology venture capitalist Marc Andreessen thinks so, praising DeepSeek as "one of the most amazing and impressive breakthroughs I have ever seen, and as open source technology, it is a great gift to the world."

But what is open source AI and why is it such an important breakthrough?

What makes DeepSeek different?

DeepSeek has disrupted the tech world. The Financial Times compared its founder, Liang Wenfeng, to the "David" who challenged the "Goliath" of the big American tech giants.

DeepSeek, based in Hangzhou and backed by Chinese hedge fund High Flyer, was founded in 2023. Its flagship DeepSeek-R1 model, launched in January 2025, has demonstrated comparable reasoning and mathematical skills to other leading AI models. But its development process has some key differences from other AI models.

Unlike OpenAI's ChatGPT and Anthropic's Claude (whose models, datasets, and algorithms are confidential), DeepSeek is open source, which means anyone can download, copy, and build on it. Its code and comprehensive technical description are shared freely and can be accessed, modified, and deployed by developers and organizations around the world.

In contrast, while Meta and Google’s models are open to anyone, they are not considered true open source models because the ways their users can apply the models are restricted by their licenses and the training datasets are not publicly available.

The United States remains the main source of AI models.

Image source: Stanford University

Lower cost, higher energy efficiency

Another key difference between DeepSeek and other AI models is the cost of building it. Data shows that it cost $5.6 million to build, which is only 10% of the cost of Meta's Llama model. This makes people question whether the huge financing needs of the AI ​​market (especially the US market) are as necessary as people initially thought. As a result, the stock price of Nvidia, which designs AI semiconductor technology, fell sharply in late January.

If AI models can be built in a shorter time and with lower energy consumption, it would be good news for the environment: the high electricity consumption associated with AI is significantly increasing carbon emissions. DeepSeek's approach could provide a more sustainable path for scaling AI.

The United States has invested more in artificial intelligence than the European Union and China.

Image source: Stanford University

How is open source AI changing the market?

As of the end of January 2025, DeepSeek has surpassed ChatGPT to become the most downloaded free app in the Apple App Store in the U.S. Many believe that its rapid rise could be an important step in the democratization of artificial intelligence, enabling small companies, startups, and individual developers to build on DeepSeek-R1.

Some argue that a more egalitarian AI-for-all approach could accelerate innovation in areas where access to cutting-edge technology is difficult and drive technological progress more broadly. Rather than focusing on building specialized models, developers could use their resources to create specialized applications, thereby unleashing the power of AI to solve real-world problems.

This goal is also supported by the World Economic Forum’s Industry AI Transformation initiative, which is exploring the challenges and opportunities presented by AI-driven innovation and has explored the transformative potential of AI across industries in a newly released white paper.

Many in the tech community believe that DeepSeek’s open source nature will help foster a collaborative environment and accelerate AI innovation, while others believe that open source models may be more trustworthy because people can query the training data.

In fact, many leading models have led to legal action against their companies due to the lack of transparency in their training data.

However, DeepSeek has also been criticized for alleged censorship issues in its answers and training data. Some governments have also banned the use of DeepSeek on the grounds of privacy issues. Ensuring transparent data processing, robust security measures and minimal review is key to building trust and maximizing technology acceptance.

The number of AI-related regulations in the United States is rising rapidly.

Image source: Stanford University

On the other hand, some believe that open source AI poses huge risks as it could be used for malicious purposes, such as developing biological weapons and facilitating the spread of misinformation and disinformation.

But in any case, the emergence of DeepSeek has made the artificial intelligence industry rethink its competitiveness. Its low-cost open source approach has overturned traditional concepts and prompted countries to reflect on the true path to success in the era of artificial intelligence.

While DeepSeek’s open source framework promotes inclusivity and transparency, it also raises important questions about data privacy, geopolitical dynamics, and security. As the technology continues to develop, how these challenges are addressed will be key to unlocking its full potential and ensuring it is used responsibly.

Charlotte Edmond

Senior Writer, Forum Agenda

<<:  How to deal with 4.3 million Note7s? Samsung's new challenge

>>:  It is rumored that BYD will acquire a new car company, it is definitely not NIO, who do you think it will be

Recommend

How to market local restaurants?

A noodle shop that doesn't engage in fancy ma...

The past and present life of a steamed bun

As a valuable heritage of Chinese traditional foo...

Super detailed business logic process of login and registration

Login and registration are reflected differently ...

Luckin Coffee community operation skills!

After Luckin Coffee was suspended and delisted, i...

The top-quality watermelon from 100 years ago is not even eaten by pigs now?

Do you still remember that a blogger once posted ...

Download the full version of the package dating method on Baidu Cloud

Introduction to Package Date This course is a cou...

How a good rice variety is cultivated

In April 2022, General Secretary Xi Jinping visit...

What are the requirements for a bidding promotion creative?

There should be at least 1-2 wildcards in the tit...

New ideas for attracting new users: active and passive user acquisition

The article mainly explains in depth and in a sim...

Android external file encryption and decryption and application practice

There is such an application scenario. When we pu...

Can you make a phone call in a vacuum? Why?

This article is based on answering similar questi...