On Nov. 30, 2022, OpenAI, an American artificial intelligence research organization, released a demo version of ChatGPT. The chatbot quickly went viral, reaching over one million users in less than a week. Just two years later, in December 2024, ChatGPT.com received approximately 4.79 billion visits.
With OpenAI’s success, American companies like Microsoft, Google, Anthropic, and Meta have built upon and developed their own AI tools. In April 2023, an EU report highlighted that 73 percent of large language models (LLMs) were being developed in the U.S. while China accounted for 15 percent.
Yet, U.S. dominance in AI development has recently been called into question.
On January 20, the China-based company DeepSeek released R1, a competitor to ChatGPT that was reportedly made for a fraction of the cost and with much less computing power. R1’s base model was developed at an estimated cost of under $6 million, far less than the over $100 million that OpenAI CEO Sam Altman stated was required to train GPT-4. This has introduced concern that the U.S. is now lagging in the AI race.
Some have branded the release of DeepSeek as a “Sputnik moment,” likening the competition between American and Chinese AI research organizations to a new Cold War. But is the release of DeepSeek R1 comparable to when the Soviet Union launched the first satellite into orbit?
The Center for Strategic and International Studies offers an alternative viewpoint, describing DeepSeek’s R1 as a “new chapter in the AI Race.” CSIS finds that DeepSeek marks an improvement in efficiency but notes that “Chinese AI models still depend on U.S.-made hardware.”
Professor Jasser of the Data Analytics department at Rollins supported this view. He compared DeepSeek’s development to modifying a car, increasing its top speed from 180 to 200 miles per hour and then making it slightly cheaper. “I took your technology and built upon it, but I still lack the knowledge of how to invent the car. That is the case of China and the U.S. The U.S. invents these technologies. China imitates and enhances them,” said Jasser.
Though China has progressed in “closing the AI gap,” Jasser maintained that the U.S. holds the advantage in AI development. “OpenAI’s o3 performs better than DeepSeek. Google released Gemini 2.0 Flash, and it is performing better than DeepSeek,” he said.
Performance comparisons aside, the release of DeepSeek has raised concerns about security and censorship. A CBS news analysis found that DeepSeek did not return results for a prompt about the 1989 Tiananmen Square protests. The platform also does not recognize Taiwan as an independent country. Still, some users have managed to bypass filters by requesting the AI to replace certain letters with numbers in its response.
This month, Taiwan banned government agencies from using DeepSeek. Italy has taken measures to block the chatbot due to privacy concerns. Other countries like South Korea, Ireland, France, Australia, and the Netherlands are considering restrictions as well.
As China competes with the U.S. for dominance in the AI market, censorship policies around generative AI may limit its global reach.
Comments are closed.