DeepSeek, Qwen, and the AI Race: A Lunar New Year Shake-Up
In last week’s newsletter, I mentioned DeepSeek—little did I know it would make such a significant impact on the stock market this week [Read more]. What a way to ring in the Lunar New Year of the Snake, a symbol of transformation in some cultures [Read more]. This week, two new AI models from China have everyone talking—DeepSeek’s R1 and Alibaba’s Qwen, the latter claiming to surpass DeepSeek’s model [Read more].
I’m fortunate to be teaching an Introduction to AI and Deep Learning course to Master of Management in Analytics (MMA) students at McGill [Program details]. While many are focused on the geopolitical implications of DeepSeek’s developments, my students are more interested in exploring how the new R1 model can be applied to various tasks. Some of the articles below come courtesy of my MMA and MBA students.
It was particularly interesting to see OpenAI's response to the DeepSeek R1 announcement. On one hand, they welcomed the update [See Sam Altman's reaction]. At the same time, they raised concerns about the unauthorized use of their data to train R1 (I know some of you might be rolling eyes right now) [Read more]. Meanwhile, OpenAI also rolled out its new reasoning model, o3-mini. While the development was announced back in December 2024, the timing of its free release this week speaks for itself [Read more].
The geopolitical dimension is becoming increasingly significant. More users are raising concerns about the safety and privacy of LLMs, while the Indian government is focusing on developing a world-class AI model to compete with China [Read more]. I'll be closely following these developments, especially as new U.S. sanctions are expected in the coming days [Read more].
Some other interesting reads, thanks to my students:
"On DeepSeek and Export Controls" by Dario Amodei
Read here"China’s DeepSeek AI Sparks Concerns Over Nvidia, OpenAI, and Export Limits" (WSJ)
Read here"DeepSeek AI and the Sovereignty Question for India" by Nadh
Read here
