DeepSeek: What is China’s new groundbreaking AI that beats OpenAI against all odds?
DeepSeek: What is China’s new groundbreaking AI that beats OpenAI against all odds?
Share:
Experts are lauding the open-source nature of the AI model. Chinese startup DeepSeek has released a new open-sourced “low cost” artificial intelligencemodel R1, rivalling ChatGPT, earning both appreciations and concerns from tech experts in Silicon Valley.
The AI company, which seems to match OpenAI’s newer 01 model in several benchmarks, claimed in a study that it spent less than $6 million to train its model compared to the hundreds of millions of dollars that American companies pour to train theirs.
After “comprehensive evaluations”, DeepSeek said its AI model “outperforms other open-source models and achieves performance comparable to leading closed-source models”. “Despite its strong performance, it also maintains economical training costs,” DeepSeek researchers wrote.
The AI startup’s achievement has come despite US sanctions denying China access to advanced semiconductors like Nvidia’s H100 GPUs, and DeepSeek refininig its algorithms by optimising less sophisticated H800 chips. Hancheng Cao, an assistant professor in information systems at Emory University, hailed the AI model as a “truly equalising breakthrough”.
DeepSeek’s model could be “great for researchers and developers with limited resources, especially those from the Global South,” Dr Cao told MIT Technology Review. The R1 app has quickly climbed to the top spot among free apps in the Apple App Store, just ahead of ChatGPT, sparking a debate on whether the Chinese firm was posing a threat to its American competitors.
Share:
More for You
Top Followed
You Might Also Like
