A Carnegie Mellon University computer science professor has lamented the “confusion” about his former student, Zhilin Yang, head of China-based start-up Moonshot AI.
Mr Yang's start-up caused consternation and controversy in the US late last week when it introduced Kimi K3, an AI language model that many analysts said rivalled and even surpassed offerings from OpenAI, Anthropic and Google.
One of the secondary but prevalent concerns raised was the fact that Mr Zhilin chose not to stay in the US to pursue his technology start-up ambitions.
Those worries and debates come as the US and China pursue a race for AI dominance and, in turn, take an increasingly adversarial view of each other.
Both countries have sought to attract and retain the best AI talent.
“I think there's a lot of confusion and misinformation about Zhilin's visa, H1B lottery, immigration, and why he left the US,” Prof Russ Salakhutdinov wrote on X. As he approached the end of his studies at CMU, Mr Zhilin had several opportunities to work for US technology companies, including Apple, which has an office in Beijing, he added.
“Zhilin was quite determined to go back and build a start-up,” he said.
“I remember him telling me that if he didn't at least try starting his own company, he would regret it for the rest of his life.”
Prof Salakhutdinov, who works in the machine learning department of Pittsburgh-based CMU and has had jobs in AI for both Apple and Meta, briefly addressed those who criticised Mr Zhilin for not starting Moonshot Ai in the US, while also acknowledging others who blamed increasingly stringent immigration policies under President Donald Trump.
“Many of my international PhD students do choose to stay in the US,” he said.
“Of course the US immigration process can be quite intimidating and uncertain, even for superstar PhD graduates from places like CMU.”
Last week as Moonshot AI's Kimi K3 started to dominate technology sector headlines and social media platforms, Mr Salakhutdinov congratulated Mr Zhilin who graduated in 2019 and is now 34.
He said that the latest Kimi AI model was a victory for the open-source AI community.
“It feels like just yesterday Zhilin was graduating from my lab at CMU,” Prof Salakhutdinov wrote.
“Not only did he complete his PhD in just four years, but he also made truly fundamental contributions to machine learning during his time at Carnegie Mellon University.”
Mohammed Soliman, technology analyst and director at the Washington-based advisory firm McLarty Associates, said that Kimi K3 was not necessarily superior to US AI models. But that may not matter, he added.
“It's close enough, even with the constraints China is working under,” he said. China has limitations due to of semiconductor manufacturing.
“The gap has closed faster than a lot of people expected, and that changes how we should think about export controls,” Mr Solimon added.
He said that the US, however, still has a major advantage.
“The US still holds an edge in efficiency and that matters a lot when you’re trying to run these systems at scale,” he said.
Kimi K3, he said, closely rivals US models, but at a cost.
“It often burns through more compute to get there,” he added.
For several years, he said, many had assumed the US would be able to maintain a permanent lead with its various companies pursuing a closed-model approach.
Mr Solimon said that with the debut of Kimi K3, that narrative will drastically change.
“They [China] release capable systems when they’re still catching up, which accelerates adoption and makes it harder for American labs to maintain a permanent lead through closed development alone,” he said.
Supporters of open-source software and AI models, where the code and technical specifications are freely available, are said to help to democratise AI and speed up popularity.
