, an assistant professor of electrical and computer engineering at University of Wisconsin–Madison, is available to comment on the Chinese AI model DeepSeek and its significance for AI development globally and in the U.S.
Some of Prof. Lee's high-level thoughts are included here. He is also available for interviews. Contact UW–Madison College of Engineering News Manager for an interview.
What’s significant about this development?
It is significant because (1) it matched one of the most advanced reasoning models we had, such as OpenAI's o1, with very novel and efficient approaches, and (2) they publicly shared the model weights so that anyone can use it freely.
Furthermore, some of the models they released are quite strong at reasoning despite being very small (runnable on the typical laptop, for instance), which was not previously possible.
What, if anything, does it mean for AI development in the US?
First, it was commonly believed that the US was far ahead of other countries, and not many people foresaw such intense international competition.
Second, due to the new efficient training paradigm they developed and the small yet powerful reasoning models they released, the scale of AI training and deployment (and consequently the demand for AI chips) could drop significantly in the short term.
In the long run, this will likely open up even more application areas and possibilities, though the full impact will take time to materialize.
What do you think are some interesting takeaways from this development?
Innovation thrives under constraints! It's very impressive that a relatively small organization with limited access to computing power achieved such a remarkable feat. As Open AI CEO , it's "legit invigorating" for all of us!
RSS