In the latest turn of events, Chinese startup DeepSeek launched its open-source AI model to rival OpenAI, Google and other US tech giants. Moreover, according to the New York Times, the company utilised significantly fewer specialised computer chips to develop the model than its US counterparts. To build DeepSeek R1, the company relied on pre-sanction of Nvidia A100 chips and less powerful alternatives, which forced the teams to find creative solutions to it.
These chips play a pivotal role in US-China technological competition. The USA imposed a trade ban on China to curb its access to Advanced AI chips in an attempt to maintain its AI dominance. However, DeepSeek’s success highlights the potential unintended consequences of these bans and has sparked conversations on the global AI landscape.
According to Meta’s Chief AI scientist, Yann LeCun’s, X (Twitter) post, “To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI.’ You are reading this wrong. The correct reading is: ‘Open-source models are surpassing proprietary ones.” He emphasised the role of the growing competence of open-source models. A similar optimism was shared by the Y Combinator CEO Garry Tan. He states the development of DeepSeek R1 will further accelerate AI adoption in the region.
Also read, ConocoPhillips Gets Sole Ownership of KBBC PSC
On the contrary, Neal Khosla, CEO of Curai, stated that the company is faking its cost to develop its AI model as a part of a larger geopolitical strategy.
DeepSeek R1: Milestone in Creativity & AI
DeepSeek R1 is a cutting-edge large language model (LLM) built for reasoning-intensive tasks such as mathematics and programming. Developed in Hangzhou, China, its creators assert that it rivals or surpasses some of OpenAI’s well-known models, like ChatGPT o1, while being significantly more cost-effective.
A key highlight of DeepSeek R1 is its efficiency. Unlike many resource-intensive AI models, it is designed to optimise memory usage and lower computational requirements, making it more accessible to researchers and developers.
