When a small Chinese society called Deepseek revealed that she had created an artificial intelligence system that could combine the main artificial intelligence products made in the United States, the news was welcomed in many circles as a warning that China was filling the gap in the race global to build artificial intelligence.
Deepseek also claimed to have built his new technology in a more effective way and with fewer computers that are difficult to hit than his American competitors, shocking a sector that had come to believe that the biggest and best IA would cost billions and billions of dollars.
But artificial intelligence experts within the Giant of Meta Technology have seen the turning point of Deepseek as something more than the arrival of an agile and new competitor on the other side of the world: it was a claim that an unconventional decision that the Meta taken almost two years ago was the right call.
In 2023, destination, with a widely criticized move, he gave his cutting -edge artificial intelligence technology after spent millions to build it. Deepseek has used parts of that technology and other artificial intelligence tools available for free on the internet through a software development method called open source.
The leaders of the leaders believe that the turning point of Deepsek Monsters that Start up now have the opportunity to innovate and compete with the giants of technology that have mainly had the artificial intelligence game field because artificial intelligence costs so much to build. It was something that the leaders were hoping that they happened when they gave their technology.
“Our Open Source strategy has been validated,” said RagaVan Srinivasan, a destination Vice President, in an interview on Tuesday. “More people have access to the technology necessary to advance things faster, the better.”
Meta is also taking a look close to the work done in Deepseek. Following the command of Meta, the Chinese company has also released its technology to the open source technological community. Meta has created several “war rooms” in which employees are the Reverse Engineering Deepseek technology, according to two people who are familiar with the effort they spoke on condition of anonymity.
The dependent destination are looking for ways to reduce the cost of the formation of its software – a term used to describe the way in which the technologies to the data learn – and apply them to the measy the information previously reported in the war halls.
Before Meta, who owns Facebook, Instagram and WhatsApp, gave its technology AI, society had focused on projects such as virtual reality. It was captured flat when Openi introduced the chatbot chatgpt at the end of 2022. Other giants of technology such as Microsoft, the nearby partner of Openai and Google were also far ahead in their artificial intelligence efforts.
(The New York Times sued Openi and its partner, Microsoft, supporting the violation of the copyright of news related to artificial intelligence systems. The two technological companies have denied the affirmations of the cause.)
By freely sharing the code that guided its AI technology, called Llama, Meta hoped to accelerate the development of its technology and attract others to build on it. Meta engineers believed that artificial intelligence experts who worked in collaboration could make more progress than the teams of experts carved within companies, as they were in Openi and the other giants of technology.
Meta could afford to do it. He made money by selling ads online, non -software AI. By accelerating the development of the AIn offered free of charge to consumers, it could attract more attention to online services such as Facebook and Instagram and sell more ads.
“They were the only large US company to adopt this approach. And it was easier for them to do this – more defensible “, Chris V. Nicholson said, an investor of the venture capital page One Ventures, who focuses on the AI technologies. Meta can offer below the cost to build it – or even give it – to attract customers and increase the sales of other services, he added.
Many in Silicon Valley said that Meta move has created a dangerous precedent because chatbots could help spread disinformation, hatred speeches and other toxic content. But Meta said that any risks were far pleased with the benefits of the open source. And most of the development of artificial intelligence, they added, had been shared through Open Source until Chatgpt has spread the companies to show what they worked on.
Now, if Deepseek’s work can be replicated, in particular his affirmation that was able to build his to the cheapest than most had thought possible – that he could offer more opportunities to multiple companies to expand what he was doing .
“These dynamics are invisible to the US consumer,” said Nicholson. “But they are extremely important.”
Yann Lecun, a first pioneer of artificial intelligence that is the main Ai scientist of Meta, said in a post on LinkedIn that people who think that the takeaway from Deepmind’s work should be that China is beating the United States during The development of the AI are gathering the situation badly. “The correct reading is:” The open source models are overcoming those owners, “he said.
Dr. Lecun added that “since their work is published and open source, everyone can profit from it. This is the power of open research. “
By the last summer, many Chinese companies had followed the command of Meta, openly open to their work. These companies included Deepseek, which was created by a quantitative commercial company called High-Flyer.
Some Chinese companies have offered “perfected” versions technology Open from companies from other countries, such as destination. But others, such as the start 01.AI, founded by a well-known investor and technologist named Kai-Fu Lee, have used parts of the goal code to build more powerful technologies.
U.S. technology experts still argue that US companies as a destination should not be open for their technologies because they were fueling the IA in China. But others say that if American companies had freely stopped their technology, the epicenter of open source development would still move to China.
At the beginning of this year, the students of the University of California, Berkeley built an artificial intelligence system that in many ways has claimed with the performance of the last Openii system. They did it by building on the top of two open source technologies released by the Giant of Chinese technology Alibaba.
“When you are in a race to build technology, the best way to compete is to share the code, strengthen the foundation and accelerate the progress rate,” said Clément Delague, CEO of Hugging Face, a company that hosts many of the world projects to open-source.