China Catching Up with US in Algorithms, Says Ex-Microsoft AI Head

• Harry Shum notes that China is catching up in AI algorithms.
• China still lags behind in AI chip technology compared to the US.
• DeepSeek demonstrates efficiency using fewer AI chips than competitors.
• The company released language models at lower costs while achieving strong performance.
• Algorithm engineering could be key for China’s competitiveness.

China’s Rise in AI Algorithms Versus Chip Technology

In the ongoing tech rivalry between the United States and China, particularly in the realm of artificial intelligence, notable observations have been made. According to Harry Shum, a prominent computer scientist and former Microsoft AI head, while it’s clear the US maintains a significant lead in AI chip technology, China is rising quickly in the algorithms arena. At an economic summit organized by the University of Hong Kong Business School, which took place on Friday, Shum pointed out that the gap in chip production is a real hurdle that can’t be surmounted in a short span of one or two years. This means there’s still a considerable distance to cover for China in chip technology.

DeepSeek as an Example of Efficiency in AI

Shum articulated the importance of computational prowess, noting that this remains a persistent challenge for many in mainland China and Hong Kong. With the current obstacles in chip technology, his advice for the country seemed straightforward: prioritize innovation in algorithm engineering. A shining example of China’s potential, he mentioned, is a start-up called DeepSeek. Remarkably, DeepSeek managed to produce outcomes that rival those from top US firms, all while utilizing a mere 10,000 AI chips. In stark contrast, leading companies like OpenAI and Google often depend on hundreds of thousands of chips, which underscores the efficiency of DeepSeek’s approach.

Impacts of DeepSeek on Global AI Competition

What’s more, DeepSeek has captured the global spotlight this year for releasing two large language models that performed on par with similar models developed in the West, yet were created with significantly lesser investment. This underscores a trend that raises eyebrows and perhaps concerns among tech giants across the globe. As the competition around AI continues to intensify, the emergence of companies like DeepSeek highlights a critical shift, indicating that algorithmic advancements may level the playing field in ways that were previously unforeseen. While chip production may be a long game, algorithm advancements could be the ace up China’s sleeve in this fierce competition.

In summary, while China continues to face significant challenges in AI chip production, the advancements in algorithm technology are noteworthy. Harry Shum emphasizes the need for China to prioritize algorithm engineering, exemplified by the success of DeepSeek. This may well be the key to reshaping the landscape of AI competition in the near future.