DeepSeek Seeks Home At Marist

A phone with the DeepSeek app open on a table. Photo by Matheus Bertelli.

Marist University students may have a new AI tool coming their way from all the way across the world. 

From the massive rise in companies like OpenAI and NVIDIA to President Donald Trump’s announcement of $500 billion to be provided for AI research, the United States is pushing to be the world leader in artificial intelligence. 

Meanwhile, China has emerged, rivaling the United States as a global power in the race for powerful artificial intelligence, with Chinese-based company DeepSeek announcing the release of their most recent AI model, DeepSeek R1.

R1’s performance is comparable to OpenAI’s o1 model, which comes as a massive shock to  Silicon Valley. While ChatGPT 4o took around $41-78 million to make and recently reached $157 billion in total funding as of Oct 2024, DeepSeek R1 took only $5.9 million.

An infamous bug with ChatGPT’s inability to count the number of times a letter appears in a word is an issue that DeepSeek seemingly does not have. According to an article by the BBC, programmers have seen DeepSeek’s R1 model perform better with coding assistance than ChatGPT. Deepseek also offers more sources when asked questions while falling short in creating poems and tables. 

“If I knew more about it, I would probably use it,” said Ethan Orchanian ‘28. 

The immediate fallout was noticeable, with OpenAI’s stock worth falling 10.93% from Jan. 24 to 28, which equates to roughly $600 billion in losses before some recovery. DeepSeek also rose to be the top free app in the United States in the iOS App Store.

“I think it’s a good thing if there are other websites and resources out there,” said an anonymous junior. 

Other students share concerns over DeepSeek's environmental impacts, with it using 87% more energy consumption per prompt than Meta’s AI due to how long the average response length is.

Others are also critical of the censorship the AI engages in, with DeepSeek avoiding information about the Tiananmen Square Massacre or any that criticizes China. Even if the AI is tricked into providing this information, it will often self-correct to delete the provided information before it completes the prompt. 

While students' opinions on DeepSeek are mixed, the views of the US government have been even less positive. Over fear of data security, much like the causes behind the temporary TikTok ban, the U.S. Navy quickly banned the usage of the app, followed by the New York State government. There have also been recent talks about a federal ban on all government devices from multiple U.S. representatives. 

“If given the opportunity, I will use it,” said Patric Daileg ‘25. “But as of now, I have no reason to.”

Whether supported or not, this new AI model is an underdog entering the competitive AI market, and university students will be on the front line of adapting it into our daily lives. As smaller companies enter the competition, the push toward innovation will be more critical and rapid than ever.