By South China Morning Post | Created at 2025-03-03 13:06:15 | Updated at 2025-03-03 20:53:23
7 hours ago
The Chinese state foundation in charge of allocating funds for basic research said artificial intelligence (AI) development should not be “mere competition of computing power”, a conclusion it reached after gathering experts to discuss the impact of DeepSeek’s models.
Lu Jianhua, deputy head of the National Natural Science Foundation of China (NSFC), said DeepSeek “represents the advanced level of current AI development” and that AI research should focus on “improving its intelligence level rather than mere competition of computing power”, according to a statement on the agency’s website on Monday.
Hangzhou-based DeepSeek took the world by storm in January after unveiling its open-sourced AI models built at a fraction of the cost incurred by its bigger US rivals. It achieved good model performance through system optimisation, sparking doubts about scaling laws, the principle that says larger training data and model parameters enhance a model’s intelligence.
In China, DeepSeek’s open-source models have kick-started a wave of AI adoption among businesses, consumers and institutions.
According to Lu, a takeaway from the success of DeepSeek was that China should “focus on core technology breakthroughs of large models, hardware and software synergy, and industry applications” so that the country can “build an autonomous and controllable ecosystem”, a slogan that refers to reducing supply dependence on the US.
